OBJECTIVES:  Develop definitions for clinically meaningful ‘within-subject change’ on the Migraine Physical Function Impact Diary (MPFID), an instrument evaluating migraine impact on physical functioning. METHODS:  Responder definitions (RDs) were developed using interim data from an observational study of episodic migraineurs (EM) who recently initiated or changed their migraine prophylaxis regimen. Subjects completed MPFID and a headache diary daily for 4 months. MPFID domain scores (Impact on Everyday Activities [EA]; Physical Impairment [PI]) range from 0–100 (higher score=greater impact). Monthly MFPID domains were averaged daily scores across migraine and non-migraine days. Differences in MPFID domain scores between months 1 and 4 were used to estimate RDs for average within-subject change using (a) multiple external anchors/indicators of change: monthly migraine days, Patient Global Impression of Change (PGIC) in overall migraine since first visit, migraine interference with daily activities, and MPFID global impact item, and (b) distribution-based analyses. RESULTS:  The sample included 78 EM subjects. EA change scores by anchor-based groups were: -4.7 for ≥50% monthly migraine reduction, -4.7 for PGIC ‘a little better’ to -3.7 for combined ‘moderately/a little better,’ -3.4 for 1–2 point change in interference, and -5.5 points for ≥20% change on MPFID global impact. PI change scores were: -3.1 for ≥50% monthly migraine reduction, -3.3 for PGIC ‘a little better’ to -2.5 for combined ‘moderately/a little better’, -2.2 for 1–2 point change in interference, and -4.0 points for ≥20% change on MPFID global impact. Distribution-based results (0.50 SD: EA=5.4, PI=5.8) supported anchor-based methods. MPFID cut-points for ‘within-subject change’ were estimated at 4–5 points for EA and 3–5 points for PI domain scores based on this study. CONCLUSIONS:  Findings suggested that change of 3–5 points for MPFID EA and PI domain scores were related to clinically meaningful change as assessed by various anchors. RDs will be confirmed using additional data.