A model for the timing of repetitive discrete motor responses is proposed, and a prediction of negative dependency between successive interresponse intervals is confirmed by data from a Morse key tapping task. The model suggests that interresponse intervals (I) are determined by the sum of a timekeeping process (C) and a delay process (D), where I = C - D_{j-1} + D_j. The mean of I is equal to the mean of C, and the variance of I is the sum of the variances of C and D. The lag one serial correlation (ρ_I(1)) between successive interresponse intervals is given by ρ_I(1) = -1/(2 + (σ_C²/σ_D²)). This result indicates that negative dependence between successive intervals does not necessarily result from feedback of temporal information from the previous interval. Instead, it could be an artifact of delays in the system subsequent to the controlled timing of a particular interval. The study also examines the relationship between interresponse interval and the variance of the timekeeper. Data from experiments with different subjects and intervals show that the lag one serial correlation (ρ_I(1)) is always in the range 0 > ρ_I(1) > -1/2. The estimates of the variances of the timekeeper (σ_C²) and response delay (σ_D²) are shown to vary with interresponse interval. The results support a two-process model for the timing of discrete motor responses, where both the timekeeper and response delay processes contribute to the variance of interresponse intervals. The study also contrasts the two-process model with other approaches to reaction time analysis, and suggests that the model provides a reasonable characterization of the formal requirements of a fixed-rate repetitive response task. The results confirm the prediction that ρ_I(1) lies within the range 0 > ρ_I(1) > -1/2 for all subjects. The study also suggests that the influence of variability in response delays plays a less significant role at longer interresponse intervals. The method derived from the two-process model for decomposing an overall interval variance into the two variances of the timekeeper and response delay processes is contrasted with other approaches to reaction time analysis. The study concludes that the two-process model provides a reasonable characterization of the formal requirements of a fixed-rate repetitive response task.A model for the timing of repetitive discrete motor responses is proposed, and a prediction of negative dependency between successive interresponse intervals is confirmed by data from a Morse key tapping task. The model suggests that interresponse intervals (I) are determined by the sum of a timekeeping process (C) and a delay process (D), where I = C - D_{j-1} + D_j. The mean of I is equal to the mean of C, and the variance of I is the sum of the variances of C and D. The lag one serial correlation (ρ_I(1)) between successive interresponse intervals is given by ρ_I(1) = -1/(2 + (σ_C²/σ_D²)). This result indicates that negative dependence between successive intervals does not necessarily result from feedback of temporal information from the previous interval. Instead, it could be an artifact of delays in the system subsequent to the controlled timing of a particular interval. The study also examines the relationship between interresponse interval and the variance of the timekeeper. Data from experiments with different subjects and intervals show that the lag one serial correlation (ρ_I(1)) is always in the range 0 > ρ_I(1) > -1/2. The estimates of the variances of the timekeeper (σ_C²) and response delay (σ_D²) are shown to vary with interresponse interval. The results support a two-process model for the timing of discrete motor responses, where both the timekeeper and response delay processes contribute to the variance of interresponse intervals. The study also contrasts the two-process model with other approaches to reaction time analysis, and suggests that the model provides a reasonable characterization of the formal requirements of a fixed-rate repetitive response task. The results confirm the prediction that ρ_I(1) lies within the range 0 > ρ_I(1) > -1/2 for all subjects. The study also suggests that the influence of variability in response delays plays a less significant role at longer interresponse intervals. The method derived from the two-process model for decomposing an overall interval variance into the two variances of the timekeeper and response delay processes is contrasted with other approaches to reaction time analysis. The study concludes that the two-process model provides a reasonable characterization of the formal requirements of a fixed-rate repetitive response task.