The purpose was to investigate the effect of variation in shock intensity on the acquisition of a black-white discrimination by 56 white rats in a situation involving shock for the correct or incorrect response, a noncorrection procedure, and 2 different training methods (free-vs. forcedchoice). Differences in outcome between training methods were negligible. Shock intensity functions showed that errors increased with increasing intensities for shock-right groups, decreased with increasing intensities for shock-wrong groups. All shock-right groups were inferior to no-shock controls, a finding in opposition to the generalization that shock for the correct response facilitates performance.