Three pigeons, previously trained to a high level of accuracy on matching-to-sample procedures, were exposed to various schedules of ratio reinforcement for correct matches. Overall accuracy was lower on fixed ratio than on regular reinforcement. There was a high incidence of errors immediately after reinforcement on fixed ratio schedules with accuracy increasing as the ratio progressed. This increase was found to be inversely correlated with the latency of the observing response to the sample. By contrast, accuracy was high throughout the ratio on a variable ratio schedule.