Prevalence of the extinction burst and its attenuation during treatment.

The team looked at 113 published data sets where extinction was used. They asked how often an extinction burst shows up.

They also checked if adding other procedures, like reinforcement for a new response, changes the odds of a burst.

Bursts happened in only 24 out of every 100 cases. That is about one in four.

When extinction was paired with other operant procedures, the rate fell to 12 percent. Used alone, it stayed at 36 percent.

Muething et al. (2024) repeated the count and landed on the same 24 percent figure. This direct replication gives you confidence the one-in-four rule still holds today.

Hatton et al. (1999) narrowed the lens to self-injury only and saw bursts or aggression in roughly half the cases. The higher number does not clash with our target paper; it simply shows that severe problem behavior makes bursts more likely.

Shahan et al. (2025) showed that giving a bigger alternative reinforcer during extinction shrinks or even erases the burst. This lab finding explains why the 1995 data saw fewer bursts when extra procedures were in play.

You can now set realistic expectations: expect a burst only about a quarter of the time, and cut that risk in half by pairing extinction with other supports such as differential reinforcement. When you treat high-risk behaviors like self-injury, plan for the higher end of the range, but still buffer the program with strong alternative reinforcement. Document the first three minutes of each session; that is when most bursts appear and when your quick response matters most.