The TIGER-3 (NCT02322281) study was initiated to compare the efficacy and safety of rociletinib, a third-generation EGFR tyrosine kinase inhibitor (TKI) that targets EGFR T790M and common EGFR-activating mutations, versus chemotherapy in patients with NSCLC who progressed on first- or second-generation EGFR TKIs. Patients with advanced or metastatic EGFR-mutated NSCLC with disease progression on standard therapy (previous EGFR TKI and platinum-based chemotherapy) were randomized to oral rociletinib (500 or 625 mg twice daily) or single-agent chemotherapy (pemetrexed, gemcitabine, docetaxel, or paclitaxel). Enrollment was halted when rociletinib development was discontinued in 2016. Of 149 enrolled patients, 75 were randomized to rociletinib (n = 53: 500 mg twice daily; n = 22: 625 mg twice daily) and 74 to chemotherapy. The median investigator-assessed progression-free survival (PFS) was 4.1 months (95% confidence interval CI: 2.6–5.4) in the rociletinib 500-mg group and 5.5 months (95% CI: 1.8–8.1) in the 625-mg group versus 2.5 months (95% CI: 1.4–2.9) in the chemotherapy group. An improved PFS was observed in patients with T790M-positive NSCLC treated with rociletinib (n = 25; 500 mg and 625 mg twice daily) versus chemotherapy (n = 20; 6.8 versus 2.7 mo; hazard ratio = 0.55, 95% CI: 0.28–1.07, p = 0.074). Grade 3 or higher hyperglycemia (24.0%), corrected QT prolongation (6.7%), diarrhea (2.7%), and vomiting (1.3%) were more frequent with rociletinib than chemotherapy (0%, 0%, 1.4%, and 0%, respectively). Rociletinib had a more favorable median PFS versus chemotherapy but had higher rates of hyperglycemia and corrected QT prolongation in patients with advanced EGFR-mutated NSCLC who progressed on previous EGFR TKI. Incomplete enrollment prevented evaluation of the primary efficacy end point.