The purpose of this study is to examine the alternate forms reliability of the AMPS (Assessment of Motor and Process Skills) (Fisher, 1997a) where alternate forms means different pairs of AMPS tasks. The participants for this study were persons selected from the AMPS database who had performed four AMPS tasks. The participants varied in age, gender, diagnosis, and level of assistance needed to live in the community. The AMPS was administered by trained and calibrated occupational therapists according to standardized procedures. The data for the 91 participants were subjected to 12 many-faceted Rasch analyses to generate ADL motor and ADL process ability measures for each task and each set of paired tasks. Repeated measures ANOVAs revealed no time effect across the four AMPS tasks. Pearson product moment correlations between Tasks 1 and 2 combined and Tasks 3 and 4 combined were r = .91 and r = .86 for the ADL motor and ADL process scales, respectively. Calculation of the standardized difference (z) revealed that no more than 8% of the participants had ADL motor or ADL process ability measures that differed significantly between observations once we accounted for real differences in a persons performance; 80% of the paired ADL motor and ADL process ability measures remained stable within +/- 0.5 logits when the participants performed two tasks. The AMPS ADL motor and ADL process scales can be used reliably in clinical practice and for research purposes.