It has repeatedly been shown that high serum leptin levels at target weight ensue from therapeutically induced weight gain in patients with anorexia nervosa (AN). It was hypothesized that elevated leptin levels may be an important factor underlying the difficulties of maintaining the target-weight in AN patients after re-feeding. The aim of this study was to examine if serum leptin levels at discharge from inpatient treatment predict renewed weight loss within 2 months after discharge and upon a 1 yr follow-up. Univariate variance analysis (ANOVA) revealed that 60% (cor. R2=0.60, P=0.002) of the variance in the BMI standard deviation score (BMI-SDS) 2 months after discharge was explained by the model consisting of the independent variables lg10 leptin levels at discharge (P=0.019) and at admission (P=0.069) and BMI-SDS at admission (P=0.002) and delta BMI between admission and discharge (P=0.047). Similarly, 60% (cor. R2=0.60, P=0.005) of the variance in BMI-SDS 1 yr after discharge was explained by lg10 leptin levels at discharge (P=0.046) and at admission (P=0.052) and BMI-SDS at admission (P=0.008) and 2 months after discharge (P=0.007) and delta BMI between admission and discharge (P=0.933). Patients with a poor outcome after 1 yr (n=9, ANCOVA, group: descriptive P=0.041), but not recovered patients (n=9, P=0.649), had lg10 leptin levels at discharge higher than those of controls when adjusted for BMI and % body fat at discharge. In conclusion, high serum leptin levels at discharge from inpatient treatment may indicate a risk for renewed weight loss and an unfavorable 1 yr outcome in AN.