Animal studies have shown an exponential increase in the risk of Borrelia burgdorferi infection after 48-72 h of deer tick attachment. Persons with tick bites were prospectively studied to determine if those with prolonged tick attachment constitute a high-risk group for infection. Ticks were identified, measured for engorgement, and assayed by polymerase chain reaction (PCR) for B. burgdorferi DNA. Duration of attachment was determined from the scutal index of engorgement. Of 316 submissions, 229 were deer ticks; 14% were positive by PCR. Paired sera and an intact tick for determination of duration of attachment were available for 105 subjects (109 bites). There were 4 human cases (3.7% of bites) of B. burgdorferi infection. The incidence was significantly higher for duration of attachment > or =72 h than for <72 h: 3 (20%) of 15 vs. 1 (1.1%) of 94 (P = .008; odds ratio, 23.3; 95% confidence interval, 2.2-242). PCR was an unreliable predictor of infection. Tick identification and measurement of engorgement can be used to identify a small, high-risk subset of persons who may benefit from antibiotic prophylaxis.