This paper presents the results of two studies on the invariance of the Profile of Mood States questionnaire across response time frames and circumstances of administration. We applied Spanish versions of the instrument to gather data from 1146 athletes. In the first study (N = 700), we tested the factor structure of the questionnaire in training sessions by using two different time frames: 'right now' (n = 350) and 'past week' (n = 350). In the second study (N = 446), we compared the factor structure of the questionnaire with data collected using the instruction 'right now' at two different circumstances: 'training' (n = 223) and 'competition' (n = 223). Data analysis was similar in both studies. We conducted multi-group confirmatory factor analyses and applied the scaled difference chi-square statistic to examine whether discrepancies in successive constrained models were significant. We observed configural equivalence between the two time frames. Furthermore, we observed metric equivalence but not scalar invariance between the different circumstances of measurement. The findings highlight the need for studies of equivalence before using a single self-report with more than one set of instructions, or under diverse circumstances. Invariance of mood scores should be examined and taken into account when interpreting individual and group mood state assessments.