Domestic hens served as subjects in a series of three experiments investigating the effects of changeover delay (COD) on concurrent variable-interval schedule performance. The first experiment involved manipulations of the relative reinforcement rates on the two schedules presented on two separate keys while holding the COD constant over a series of COD durations from no COD to 15-sec. Sensitivity to reinforcement as measured by a in the generalised matching law was, for both response and time measures, less than unity and least at no COD. Generally it increased with the introduction of any COD but did not change systematically with COD length. Responding during the COD was found to be insensitive to changes in reinforcement rate ratios. The second experiment manipulated the stimuli presented on the main key during the COD (CO-key procedure). Adding distinctive key-colours during the COD increased sensitivity to reinforcement while the addition of blackout decreased time (but not response) sensitivity. Both effects carried over into following (non-stimulus) conditions but this did not survive a change in the schedules. The third experiment held the schedule values constant (two-key procedure) while the COD values were varied from no COD to 12-sec (thrice). Measures of behaviour changed with COD value although generally the a values were similar to those of the first experiment. Overall the changes were recoverable although individual point estimates of a were shown not to be reliable. The findings from the three experiments support the use of the generalised matching law to describe performance on concurrent schedules of reinforcement. They also question the common use of changeover delays in concurrent schedules in that they appear to do more than simply separate the schedules in time.