Background: This phase III randomized trial compared pemetrexed 500 mg/m2 (P500) with pemetrexed 900 mg/m2 (P900) to determine whether higher dosing benefits non-small-cell lung cancer (NSCLC) patients as second-line therapy. Patients and methods: Patients with locally advanced or metastatic NSCLC, previously treated with platinum-based chemotherapy, were randomly assigned to receive i.v. P500 or P900 every 3 week. Results: Accrual was terminated with 588/600 patients enrolled because an interim analysis indicated a low probability of improved survival and numerically greater toxicity on the P900 arm. P900 patients were permitted to continue treatment at P500. No statistical difference was observed between the treatment arms (P500 versus P900) for median survival {6.7 versus 6.9 months, hazard ratio [HR] = 1.0132 [95% confidence interval (CI) 0.837-1.226]}, progression-free survival [2.6 versus 2.8 months, HR = 0.9681 (95% CI 0.817-1.147)], or best overall tumor response [7.1% versus 4.3% (P = 0.1616)]. The incidence of drug-related grade 3/4 toxicity was typically <5% on both treatment arms, but was numerically higher on the P900 arm for most toxicity categories. Conclusions: P900 did not improve any efficacy measure over P500. P500 i.v. every 3 week remains the standard pemetrexed dose for second-line treatment of platinum-pretreated advanced NSCLC.