OBJECTIVE To determine variables that predict treatment failure after methotrexate (MTX) treatment of ectopic pregnancy. METHODS Retrospective cohort study. METHODS Canadian teaching hospital. METHODS Sixty patients diagnosed with and treated for ectopic pregnancy. METHODS A single dose of methotrexate (50 mg/m(2)) by i.m. injection. METHODS Resolution of serum beta-hCG or clinical evidence of treatment failure. RESULTS Treatment failure was observed following methotrexate administration in 65% of cases when initial beta-hCG was >4000 IU/L, but in only 7. 5% of patients when serum beta-hCG was <4000 IU/L (OR = 52.06, 95% CI 4.88-555.56). Patients who presented with pelvic pain without tenderness had treatment failure 56% of the time versus only 17% in those without pain (OR = 9.20, 95% CI 1.02-82.60). Treatment failure also occurred in 53% of patients presenting with vaginal bleeding versus 16% without bleeding (OR = 6.18, 95% CI 0.73-51.93). CONCLUSIONS Methotrexate should not be used to treat ectopic pregnancy when initial beta-hCG is >4000 IU/L. Caution should also be exercised in using methotrexate for ectopic pregnancy when the patient presents with bleeding or pain even without tenderness.