The objective of this study was to assess variability in symptoms and physical capabilities in chronic fatigue syndrome (CFS) participants both before and after a graded exercise intervention. Sixty-one CFS subjects participated in a 12-week randomized controlled trial of either graded exercise (n =32) or relaxation/stretching therapy (n = 29). Specific physiological, psychological and cognitive variables were assessed once weekly over a four-week period both prior to and after the intervention period. All scores were assessed for reliability using an intraclass correlation coefficient (ICC). Apart from mental and physical fatigue, baseline ICC scores for all variables assessed were moderately to highly reliable, indicating minimal variability. Baseline scores for mental and physical fatigue were of questionable reliability, indicating a fluctuating nature to these symptoms (R1 = 0.64 and 0.60, respectively). Variability in scores for mental fatigue was reduced after graded exercise to an acceptable classification (R₁ = 0.76). Results from this study support a variable nature to the symptoms of mental and physical fatigue only. Consequently, in order to more accurately report the nature of mental and physical fatigue in CFS, future studies should consider using repeated-measures analysis when assessing these symptoms. Graded exercise resulted in the reclassification of scores for mental fatigue from questionable to acceptable reliability.