Four types of stimulus-food contingencies were compared for effectiveness in producing and maintaining trial-stimulus pecking in pigeons. The four trial-stimulus contingencies were: (a) fixed-length trials contiguous with food, (b) variable-length trials contiguous with food, (c) fixed-length trials with fixed-length trace intervals preceding food, and (d) variable-length trials with variable-length trace intervals preceding food. In all cases the inter-trial interval was variable with a 30-sec. mean and the trial stimulus was 5 sec. or 5 sec. on the average. Trial-stimulus pecking was rapidly acquired when the trial stimulus was contiguous with food, whereas the trace procedures produced pecking in only a few pigeons and then only after considerable training. Neither variability in the trial-stimulus length nor in the trace interval reliably affected acquisition or maintenance.