How to Calculate Time Constant from a Graph

Updated 10 November 2026 · Standard formula, shown below

This guide explains how to determine the time constant (τ) of a first-order system from its response graph, such as a charging or discharging capacitor, a cooling object, or a simple RC/RL circuit. You will learn to identify the time constant using the graphical method (the 63.2% rule) and the initial slope method. No prior knowledge beyond basic graph reading is required.

What You’ll Need

A graph of the system’s response over time (e.g., voltage vs. time, temperature vs. time).
A ruler or straight edge.
A pencil (for marking points on the graph).
Optional: a calculator for verifying the 63.2% value.

Step-by-Step

There are two common graphical methods. The first (63.2% rule) is the most reliable for real-world data; the second (initial slope) is quick but less accurate.

Method 1: The 63.2% Rule (Exponential Response)

Identify the final (steady-state) value. Look at the graph where the response levels off after a long time. This is the value the curve approaches asymptotically. Call this value V_final (or T_final for temperature, etc.). For a discharging curve, the final value is typically 0.
Determine the initial value. If the curve starts at a non-zero value (e.g., a capacitor initially charged to 5 V), note this as V_initial. For a charging curve starting from 0, this is simply 0.
Calculate the target value. For a charging curve (from 0 to V_final): the time constant is the time to reach 63.2% of the final value. So, target = 0.632 × V_final.
For a discharging curve (from V_initial to 0): the time constant is the time to fall to 36.8% of the initial value. So, target = 0.368 × V_initial.
For a curve that starts at V_initial and rises to V_final (e.g., heating from 20°C to 100°C): the time constant is the time to reach a value that is 63.2% of the way from the initial to the final. So, target = V_initial + 0.632 × (V_final − V_initial).
Locate the target value on the y-axis. Draw a horizontal line from this value across the graph.
Find where that line intersects the curve. Mark this intersection point.
Drop a vertical line to the x-axis. The x-coordinate of this intersection is the time constant τ. Read the time value from the x-axis.

Method 2: Initial Slope (Tangent) Method

Draw a tangent line to the curve at the initial point (t=0). Use a ruler to draw a straight line that just touches the curve at the start and follows the curve’s initial steepness.
Extend the tangent until it crosses the final value line. Draw a horizontal line at the final steady-state value. The tangent line will intersect this horizontal line.
Read the x-coordinate of that intersection. This is the time constant τ. This works because the initial slope of an exponential is exactly 1/τ, so the tangent reaches the final value at t = τ.

Common Mistakes

Using the wrong target percentage. For a charging curve, do not use 50% or 90%. The 63.2% rule is specific to first-order systems. For discharging, it is 36.8%.
Confusing initial and final values. If the curve starts at 5 V and decays to 0, the target is 0.368 × 5 = 1.84 V, not 0.632 × 5.
Reading the x-axis incorrectly. Ensure you are reading the time axis, not the sample number or index.
Using a tangent that is not actually tangent. The tangent must touch the curve at exactly one point at t=0; if drawn loosely, the result will be off.
Assuming the graph is a perfect exponential. Real data may have noise or non-ideal behavior. The 63.2% method is more robust to small deviations than the tangent method.

Tips

Always check that the curve actually reaches a plateau. If it is still rising at the end of the graph, you may be reading a transient that hasn’t settled. Extend the graph or use a longer measurement.
For noisy data, draw a smooth best-fit curve through the points before applying the method.
Use the 63.2% method for accuracy; use the tangent method as a quick sanity check. If they disagree by more than 20%, re-examine your graph.
If the graph is on a logarithmic scale, the methods above do not apply directly. Convert to a linear scale first.
Remember that the time constant has units of time (seconds, minutes, etc.). If your x-axis is in milliseconds, τ will be in milliseconds.

Troubleshooting

The curve does not start at t=0. If the graph starts at a time offset, subtract that offset from your reading. For example, if the curve begins at t=2 ms and you find the intersection at t=7 ms, the time constant is 5 ms.
The curve is not exponential. If the graph shows a linear rise or an S-shape, the system is not first-order. The 63.2% rule will not give a meaningful time constant. In that case, you may be dealing with a second-order system or a different model.
The final value is not clear. If the curve is still rising slowly, estimate the final value by extending the trend. Alternatively, use the tangent method, which does not require a precise final value—but it is less accurate.
You get a negative time constant. This usually means you chose the wrong target (e.g., using 63.2% for a discharging curve). Recalculate.
Your reading seems too large or too small. Check the units on both axes. A common mistake is mixing milliseconds and seconds.

Frequently Asked Questions

Why is it 63.2% and not 100%? In an exponential approach, the value never actually reaches the final value in finite time. After one time constant, it has covered 63.2% of the remaining gap. After 5τ, it is within 0.67% of the final value, which is why we often consider 5τ as the settling time.
Does this method work for any first-order system? Yes, as long as the response is exponential. This includes RC circuits, RL circuits, thermal systems, and many chemical or mechanical processes.
What if my graph is inverted (e.g., decays from 10 to 2)? The same logic applies. The time constant is the time to go 63.2% of the way from the initial to the final value. So target = 10 + 0.632 × (2 − 10) = 10 − 5.056 = 4.944. Find where the curve crosses that value.
Can I use the 63.2% rule if the curve has noise? Yes, but first smooth the data by drawing a best-fit line through the points. The intersection will be approximate, which is acceptable.
What is the difference between the time constant and the half-life? The half-life is the time to reach 50% of the final value. For a first-order system, half-life = 0.693 × τ. If you prefer, you can measure the half-life and multiply by 1.443 to get τ.
Why does the tangent method work? The derivative of an exponential decay e^−t/τ at t=0 is −1/τ. The tangent line at t=0 has slope −1/τ, so it reaches zero (or the final value) at exactly t = τ. This is a geometric property of exponentials.

Method: the standard formula for this calculation, rounded as each result states. See how the calculators work or report an error.