This study focused on the assessment of the structural validity of the 28 item adapted Teacher Assessment for Learning Questionnaire (TAFL-Q) in the Kenyan context. The sample consisted of 180 mathematics teachers drawn from 50 secondary schools. Confirmatory factor analysis was used to confirm the hypothesised two-factor model of the original questionnaire. The analysis resulted in a poor fit to the Kenyan sample (CMIN/DF = 2.643, RMSEA = 0.096, SRMR = 0.085, TLI = 0.653, CFI = 0.680). The factor structure was then explored using principal components analysis with Varimax Kaiser Normalisation rotation. Nine items were eliminated and a final analysis of the remaining 19 items resulted in six factors with a total explained variance of 66%. The emergent factors were further assessed through convergent and discriminant validity and the results were within the acceptable range. The six-factor model also fitted well to the questionnaire data of the Kenyan sample. The findings suggest the need for further development of the teacher assessment for learning instruments that can be applied in measuring the teachers' perceptions in the Kenyan context and sub-Saharan Africa at large.