The present work aimed to investigate the effects of contingent and "non-contingent" negative reinforcement on varying behavior (Experiment I), as well as the effects of two extinction procedures (Experiment II), on behavioral variability. In Experiment I, four experimentally naive male rats were exposed to discrete trial avoidance contingencies (Lag 1 and Aco), according to the following design: Lag 1/Aco/Lag1/Aco. The general avoidance procedure included the presentation of 0.5-s, 0.4-mA electric shocks (US) on fixed time (FT) in the presence of a light (CS). Responses in the presence of the CS turned off the CS, sounded a 0.5-s, 10-dB tone, and avoided the next US programmed to be delivered. If the avoidance response was not emitted, the US was presented at the end of the FT, the CS remained unchanged, and a new trial was immediately initiated. Under the Lag 1 contingency, three-response sequences that differed from the last sequence emitted at the session avoided the US. Under the Yoke contingency, US cancellation was contingent on the emission of a three-response sequence, variable or not, following the same reinforcement distribution obtained in the last Lag session.