Abstract Background There are substantial differences in hospital referrals between general practitioners (GPs); however, there is little research on the consequences for patient safety and further healthcare use. Objective To investigate associations between out-of-hours GP characteristics, unplanned hospital admissions, and patient safety. Methods This cohort study included all Norwegian out-of-hours services contacts from 2008 to 2016, linked to registry data on patient characteristics, healthcare use and death, and GP age, sex, specialist status, out-of-hours service experience, and prior admission proportion. We estimated the impact from GP characteristics on (i) immediate unplanned hospital admissions for “all conditions,” (ii) immediate unplanned hospital admissions for “critical conditions,” (iii) 30-day unplanned hospital admissions, (iv) 30-day hospital costs, and (v) 30-day risk of death. To limit confounding, we matched patients in groups by age, time, and location, with an assumption of random assignment of GPs to patients with this design. Results Patients under the care of older and male GPs had fewer immediate unplanned hospital admissions, but the effects on cumulative 30-day unplanned hospital admissions and costs were small. The GPs’ prior admission proportion was strongly associated with both immediate and 30-day unplanned hospital admissions. Higher prior admission proportion was also associated with admitting more patients with critical conditions. There was little evidence of any associations between GP characteristics and 30-day risk of death. Conclusions GPs’ prior admission proportion was strongly associated with unplanned hospital admissions. We found little effects on 30-day mortality, but more restrictive referral practices may threaten patient safety through missing out on critical cases.