OBJECTIVE: To ascertain whether a change in co-amoxyclav (A/C) market share was associated with changes in patient outcomes in otitis media, sinusitis, lower respiratory infection and acute exacerbation of chronic obstructive airways disease. METHODS: An integrated database of computerised records from four general practices, based in 3 Australian States was collated for the period mid 1994 to mid 1998. The database consisted of 34,242 patients and 318,234 notes. Conditions of interests and identifiable patient outcomes were sought. After identification of the change in A/C market share using segmented regression, a time series analyses was completed to define the relationship between the A/C market share and rate of patient outcomes per encounter. The A/C and all antibiotic market shares, the number of episodes of care (EOC) for the chosen conditions, the antibiotics prescribed and the patient casemix were measured. The patient outcomes chosen included hospitalization, referral to a specialist, CT scan of sinus and of chest, chest xray, any pathology tests, spirometry, bronchoscopy and sinoscopy. RESULTS: In total there were 11,378 EOC for all specified problems. 1082 people had been given A/C for at least one EOC and 5879 other antibiotics. The casemix profile for both groups was similar. The total number of negative outcomes were 1873, comprising 126 hospitalizations, 351 referrals, 794 radiology investigations, 561 pathology tests and 41 other investigations. The time series analyses revealed that for every 1% reduction in A/C market share, there was a 0.5 % increase in negative outcomes, with three month lag. The 3-month lag was the most significant (p <0.0001). The decrease in A/C market share was related to an increase in weighted costs of negative outcomes (p = 0.0024). CONCLUSIONS: A statistical, temporal association was found between a fall in A/C market share among the study GPs and a rise in negative patient outcomes.