The objectives of the present study were to examine the accuracy of older people's reports of health services resource use after discharge from acute care compared with information from healthcare providers. Paired data were obtained from health providers and a consecutive sample of hospitalised patients (aged 60 years or over) enrolled in a randomised controlled trial of hospital-at-home versus usual acute care. Retrospective reports of the use (yes/no) and number of patient hospital admissions, and community nursing, physiotherapy, health visiting and general practitioner (GP) services were obtained between baseline, and 4- and 12-week follow-ups, although the recall period varied for different resources. The comparability of paired reports was examined by using crude and chance-corrected agreement, and by testing for systematic differences in the distribution of paired responses. Out of 219 patients enrolled in the trial, 190 and 185 patients provided data at 4 (87%) and 12 weeks (84%), respectively. Crude agreement was over 72% (range = 42-93%), and chance-corrected agreement was moderate or good (kappa coefficients from 0.23 to 0.71) for 11 out of 12 comparisons. Systematic differences in the pattern of paired responses were observed for seven comparisons. Patients more often reported receiving a routine GP home visit (by 12 weeks), whilst health providers reported statistically significantly (P < 0.05) more patients admitted to hospital (12 weeks) and more admissions per person, requesting a GP home visit (12 weeks) or surgery consultation, and having district nursing (4 and 12 weeks) and physiotherapy (4 weeks only). The data indicate that patients tend to underestimate resources used compared with health providers over relatively short time frames.