Study design: Observational study. Objectives: To report the intra-rater (one rater), inter-rater (two raters) and inter-session (one subject, sessions 1-5) reliability of lower extremity modified Ashworth scale (MAS) scores among patients with chronic spinal cord injury (SCI). Setting: Tertiary Academic Rehab Centre in Toronto, Canada. Methods: MAS scores of 20 subjects with chronic SCI (C5-T10 AIS A-D412 months) were recorded for the hip abductors and adductors, knee flexors and extensors, and ankle plantar and dorsiflexors. MAS scores were assessed by two blinded raters (A and B) at the same time of day, weekly for 5 weeks using standardized test positions, a one-cycle per second metronome, with ratings recorded on the second cycle. MAS score reproducibility [intra-rater, inter-rater] were calculated using Cohen's Kappa. Intraclass correlation coefficients (ICCs) were calculated to determine inter-session (trials 1-5) reliability; Kappa values X0.81 and ICC values X0.75 were desired. Results: Intra-rater reliability was fair to almost perfect (0.2oko1.0) and differed between raters. Inter-rater reliability was poor-to-moderate (ko0.6) for all muscle groups. Inter-session reliability for a single rater was fair-to-good (0.4oICCo0.75) for all muscle groups. Conclusions: MAS was not reliable as an intra-rater tool for all raters, and showed poor inter-rater and modest inter-session reliability. MAS has inadequate reliability for determining lower extremity spasticity between raters (inter-rater) or over time (inter-session). It is recommended that the rehabilitation science community seek alternative measures for quantifying spasticity.