Importance Skin cancer screening may improve melanoma outcomes and keratinocyte carcinoma morbidity, but little is known about the feasibility of skin cancer training and clinical skin examination (CSE) by primary care practitioners (PCPs) in large health care systems. Objective To assess the association of skin cancer training and screening by PCPs with dermatology referral patterns and rates of skin biopsies. Design, Setting, and Participants In this pilot interventional study performed at the Veterans Affairs Palo Alto Health Care System, patients 35 years or older scheduled for an annual health habits screen in the PCP general medicine clinics were studied. Interventions Six PCPs underwent Internet Curriculum for Melanoma Early Detection (INFORMED) training in May 2015, and 5 screened patients during the following 14 months. Main Outcomes and Measures Proportion of dermatology referrals, subsequent skin biopsies, and PCP diagnostic accuracy for skin cancer or precancer compared with dermatologist diagnosis were assessed in screened patients 14 months before the intervention (February 18, 2014, through April 30, 2015) and after the intervention (June 18, 2015, through August 30, 2016). Results Among 258 patients offered screening (median age, 70 years; age range, 35-94 years; 255 [98.8%] male), 189 (73.3%) received CSE and 69 (26.7%) declined. A total of 62 of 189 patients (32.8%) were referred to a dermatologist after intervention: 33 (53.2%) for presumptive skin cancers and 15 (24.2%) for precancers. Nine of 50 patients (18.0%) evaluated in dermatology clinic underwent biopsy to exclude skin cancer. Correct diagnoses were made by PCPs in 13 of 38 patients (34.2%; 4 of 27 patients [14.8%] diagnosed with skin cancers and 5 of 11 patients [45.5%] diagnosed with actinic keratoses). Comparison of all outpatient visits for the 5 main participating PCPs before vs after intervention revealed no significant differences in dermatology referrals overall and those for presumptive skin cancer or actinic keratoses, skin biopsies, or PCP diagnostic accuracy with the exception of significantly fewer postintervention dermatology referrals that lacked specific diagnoses (25 [1.0%] vs 10 [0.4%],P = .01). Conclusions and Relevance This pilot study suggests that PCP-based skin cancer training and screening are feasible and have the potential to improve PCP diagnostic accuracy without increasing specialty referrals or skin biopsies. Additional studies comparing screening rates, specialty referrals, and patient outcomes in trained vs untrained PCPs are needed before screening is widely implemented in large health care systems in the United States.