To read an OTDR trace, follow the line from left to right. The horizontal axis is distance along the fiber. The vertical axis is the backscattered power in dB. A steady downward slope is normal fiber attenuation, a sudden drop with no spike is a splice or bend, a spike followed by a drop is a connector or other reflective event, and a big spike followed by noise is the end of the fiber. Once you can name those four shapes, the rest of OTDR trace analysis is mostly measuring them carefully and knowing when the instrument is fooling you.
This guide walks through each part of an OTDR graph, then gives you a repeatable routine for interpreting results, a worked example and the traps that catch new techs: ghosts, gainers and dead zones.

Key takeaways
- X axis = distance (km, m, ft or kft). Y axis = relative power in dB. Steeper slope = higher loss per km.
- Non-reflective events (fusion splices, bends) are steps down. Reflective events (connectors, mechanical splices, breaks, the fiber end) are spikes.
- A loss that is much larger at 1550 or 1625 nm than at 1310 nm points to a macrobend. A splice loses about the same at every wavelength.
- A splice reading from one direction can be wrong, even negative. Average both directions for the true loss.
- Always use a launch cable, and a receive cable if you need the far-end connector. Without them, the first and last connectors can't be measured.
What is an OTDR trace?
An OTDR (optical time-domain reflectometer) fires short, powerful laser pulses into a fiber and records the light that comes back over time. Two things come back. The first is Rayleigh backscatter, a weak, continuous return from every point along the glass. The second is Fresnel reflections, strong returns from discrete events such as connectors, mechanical splices, breaks and the cleaved end. The instrument converts time into distance using the fiber's group index of refraction (IOR). Because the light makes a round trip, the OTDR halves both the time and the loss, so the graph shows one-way distance and one-way dB.
The result is the trace, sometimes called a "shot" or "signature". It is usually saved as a .sor file. If you received a trace and can't open it, start with our guide on how to open a .sor file on any computer.
How to read an OTDR graph: the axes
Horizontal axis: distance
The X axis is distance from the OTDR port, or from the end of the launch cable if a user offset is set. Distance comes from time of flight: distance = (speed of light × time) ÷ (2 × group index). Light travels about 0.204 m per nanosecond in glass, so the IOR matters. An IOR error of 0.001 moves every event by roughly 0.07%, which is about 68 m on a 100 km span. Remember too that cable is shorter than the fiber inside it. Loose-tube fiber runs in a helix with extra length, so the OTDR's fiber distance is typically around 1% longer than the sheath distance. Allow for that before you send a crew to dig.
Vertical axis: relative power in dB
The Y axis is the backscattered power level in dB on a logarithmic scale. The absolute number at the top doesn't matter much. Differences do: the height of a step is loss, and the height of a spike relates to reflectance. Different viewers apply different offsets to the same raw data, so the absolute dB scale can differ between programs even when the losses agree.
The parts of an OTDR trace, left to right

1. Launch spike and dead zone
The trace starts with a large reflection from the OTDR's own front connector. The receiver is briefly overloaded, so for the first few meters nothing can be measured. That is the dead zone. It is the reason for a launch cable (pulse suppressor): a known length of fiber, typically hundreds of meters for single-mode, that lets the trace recover before the first connector of the link under test.
2. Backscatter slope: fiber attenuation
Between events, the trace is a straight line sloping down. The slope is the attenuation in dB/km. Typical single-mode G.652.D fiber reads about 0.32–0.35 dB/km at 1310 nm and 0.18–0.20 dB/km at 1550 nm. If a section is much steeper than its neighbors, suspect a different fiber type, a stressed cable or an analysis setting.
3. Non-reflective events: splices and bends
A fusion splice appears as a small step down with no spike. Good fusion splices are often 0.02–0.05 dB, which can be close to the noise. A macrobend also looks like a non-reflective step. The way to tell them apart is wavelength: bend loss grows sharply at longer wavelengths, so compare 1310 and 1550 nm (or 1625 nm).
4. Reflective events: connectors and mechanical splices
A mated connector pair shows a spike, from the reflection at the glass interface, followed by a step, which is its insertion loss. The spike height relates to reflectance. A UPC connector is typically around -50 dB or better, and an APC around -60 dB or better. A cleaved glass-air end reflects about -14 dB. Dirty or damaged connectors show taller spikes and bigger steps. Each reflective event also creates its own short dead zone after it.
5. Splitters
In PON networks a passive splitter shows as a large non-reflective step. Ideal values are about 3 dB per doubling: roughly 15 dB for 1:32, and typically 16.5–17 dB once real excess loss is included. Events right after a splitter are hard to resolve.
6. End of fiber and the noise floor
The far end usually shows a final reflection (bigger if the end is cleaved or open, smaller if it is crushed or angled), after which the trace drops into the noise floor, a ragged band of random signal. Anything to the right of that point is noise, not fiber.
OTDR event types at a glance
| Trace shape | Likely event | Typical values | How to confirm |
|---|---|---|---|
| Small step, no spike | Fusion splice | 0.02–0.1 dB | Same loss at 1310 and 1550 nm |
| Step, no spike, bigger at 1550 | Macrobend | Varies | 1550/1625 loss much higher than 1310 |
| Spike + step | Connector pair | 0.1–0.5 dB, -40 to -65 dB | Location matches a patch panel |
| Spike + step (small) | Mechanical splice | 0.1–0.3 dB | Location matches a closure |
| Large step | Splitter | ~3.5 dB (1:2) to ~17 dB (1:32) | Matches PON design |
| Spike, then noise | End or break | Reflectance near -14 dB if cleaved | Distance vs expected length |
| Spike with no loss | Ghost (echo) | No real step | Sits at a multiple of an earlier event's distance |
How to interpret OTDR results: a 7-step routine
- Check the setup first. Read the wavelength, pulse width, range, averages and IOR in the header. A range shorter than the fiber, or the wrong IOR, makes everything downstream suspect.
- Find the launch and receive cables. Identify the first connector of the link at the end of the launch cable. Note whether the far-end connector is visible behind a receive cable.
- Check the total length. Does the end event match the expected route length, allowing for helix factor and slack?
- Walk the events. Match each event to a known point (panel, closure, splitter) using the as-built or splice plan. Unexplained events are your faults or your artifacts.
- Measure loss properly. For splices and connectors, use the LSA (least-squares) method. It fits lines on either side of the event and removes the fiber attenuation between cursors. Use 2-point loss for section loss.
- Compare wavelengths. Overlay 1310 and 1550 nm. Steps that grow at 1550 nm are bends.
- Compare directions. For splice acceptance, average A→B and B→A. One-way readings can be off by a few tenths of a dB.

Worked example: reading a 12.5 km single-mode trace at 1550 nm
Suppose the trace shows these events, measured from the end of a 1 km launch cable:
- 0.0 km: spike and 0.3 dB step. This is the first connector, so the launch cable did its job.
- Slope from 0.5 to 3.0 km: the level falls from -10.40 dB to -10.90 dB. That is 0.50 dB over 2.5 km = 0.20 dB/km, normal for G.652.D at 1550 nm.
- 3.2 km: a 0.05 dB step with no spike. A fusion splice, comfortably inside a 0.3 dB TIA-style limit.
- 7.8 km: a 0.60 dB step with no spike. At 1310 nm the same event reads only 0.12 dB. The big wavelength difference says macrobend, probably a tight loop in a closure or a pinched cable, not a bad splice.
- 12.5 km: a reflective spike, then noise. The fiber end, and the length matches the design.
Decision: the splice is fine. The 7.8 km event needs a site visit to find and relieve the bend, followed by a re-test. You don't "fix" it in the report. The total link loss can then be checked against the FOA loss-estimation method or the project's own budget.
OTDR trace explained: the traps
Ghosts
Highly reflective connectors on short cables bounce light back and forth, and the OTDR shows ghosts: reflective spikes with no loss, at multiples of an earlier event's distance. The FOA gives an example of a 40 m cable producing a ghost at 80 m that looked like the fiber end. Clean the connectors, use APC connectors or index-matching gel on the launch cable end, or change the pulse width and range. A ghost often appears after the real end, in the noise.
Gainers
Sometimes a splice reads as a gain. Light is not being amplified. The two fibers have different backscatter coefficients (a mode-field diameter mismatch), so one direction under-reads the loss and the other over-reads it. The true loss is the average of both directions.
Dead zones hiding events
After every reflective event there is an event dead zone, where a second reflection can't be separated, and a longer attenuation dead zone, where a splice's loss can't be measured. Best-case vendor specs at the shortest pulse are often quoted around 0.5–1 m (event) and 2.5–4 m (attenuation), but check your instrument's datasheet and test conditions. Two connectors a few meters apart at a long pulse width look like one event.
Pulse width choice
Each nanosecond of pulse covers roughly 0.1 m of two-point resolution: 10 ns ≈ 1 m, 1 µs ≈ 100 m. Short pulses resolve closely spaced events but don't reach far. Long pulses reach far but blur events together. Start short for premises and short links, and medium for long outside plant.
Missing events
If a splice you know exists isn't in the event table, it may be below the OTDR's loss threshold (often set somewhere around 0.02–0.05 dB). The trace still shows it, but auto-analysis didn't report it. Re-analyze with a lower threshold, or measure it manually.
How to read an OTDR trace report
Most reports show the trace plot, an event table and a summary. The event table lists each event's number, distance, type (reflective or non-reflective), loss, reflectance, the attenuation of the section before it and the cumulative loss. The summary gives total link loss and, often, optical return loss (ORL). ORL is the total power reflected back from the whole link, expressed as a positive number where higher is better. Reflectance is per event and negative, and closer to zero is worse. Pass/fail flags come from thresholds the operator set, so confirm they match the project spec.
How SOREditor helps with OTDR trace analysis
SOREditor opens any SOR v2 file in the browser and lets you study the trace without vendor software. For trace reading in particular:
- Zoom into events to see the LSA fit windows, peak height and noise around each splice or connector.
- Re-measure any event from the trace with LSA or 2-point loss, and see the stored value next to the measured one.
- Use markers A–B to measure section loss and slope between any two points.
- Overlay a reference .sor, such as the as-built trace or the other wavelength, to spot changes and bends.
- Export the event table or trace points to CSV, or the plot to PNG, for a training deck or report.
The analysis tools are for understanding and documenting what the fiber did. If the trace shows a bad splice or a bend, the fix happens in the field: clean, re-splice, re-test. See the FAQ for supported files, or the SOREditor homepage for the full feature list.

Conclusion
Reading an OTDR trace comes down to four shapes (slope, step, spike, end) plus a little skepticism: check the setup, compare wavelengths, compare directions and watch for ghosts. Practice on real files from your own jobs. You can open a trace in SOREditor for free and step through every event with the zoom and marker tools. For more background, the FOA's OTDR page covers the physics in depth.

FAQ
What does the slope of an OTDR trace mean?
The slope between events is the fiber's attenuation in dB/km. Single-mode fiber typically reads about 0.33 dB/km at 1310 nm and 0.19 dB/km at 1550 nm.
What is the difference between a reflective and a non-reflective event?
A reflective event, such as a connector, mechanical splice, break or open end, shows a spike because light reflects off a glass-air or glass-gap interface. A non-reflective event, such as a fusion splice or bend, is only a step down in level.
Why does my OTDR show a splice with negative loss?
That is a gainer. The fibers on each side have different backscatter coefficients. Test from both ends and average the two readings to get the real splice loss.
How do I tell a bend from a bad splice on an OTDR?
Compare wavelengths. A macrobend loses much more at 1550 or 1625 nm than at 1310 nm, while a splice shows about the same loss at each wavelength.
What is the noise at the end of an OTDR trace?
It is the noise floor. After the fiber end there is no backscatter left, so the OTDR only records random noise. Events shown there are usually ghosts.