The studies aimed to determine the referral rate, the characteristics of patients referred, the reasons for referral and, in a sub-sample, its outcome. Study I recorded data on 602 patients referred to hospital from 13 rural clinics; follow-up interviews were conducted with a sub-sample of 57 referred patients. Study II interviewed 97 consecutive outpatients seeking care at Meru District Hospital. Both studies were carried out in a rural area of Meru district in central Kenya. The 602 referred cases represent between 0.3 and 1.8% of all new patients at the respective clinic with a mean of 0.5%, or 8.7 referrals annually per 1000 catchment area population. More than half of the referrals were for treatment only while 13% were for investigation, 63% were considered urgent, and 85% were referred in writing. Of the 57 subsequently interviewed, all except one had complied, and 52 patients, mostly with infections, injuries or obstetric disorders, reported improvement at the time of interview. The second study found that only 19 (20%) of the 97 outpatients had been referred (II in writing and 8 verbally). Rates of formal clinic-to-hospital referral were very low. Reasons included poor access to transport; patient inability to pay user fees, transport outlays, and other expenses; time constraints in emergency situations; and lack of feedback information from hospital to clinic. Verbal referral was common. Further research should focus on the selection of patients for referral, on follow-up of cases referred, on the communication between institutions involved, and on options for the financing of emergency transport to hospital.