Background: NALA (ClinicalTrials.gov NCT01808573) is a multinational, randomized, open-label, phase III trial of neratinib (an irreversible pan-HER tyrosine kinase inhibitor [TKI]) + capecitabine (N+C) vs lapatinib (a reversible dual TKI) + capecitabine (L+C) in patients with stage IV HER2+ metastatic breast cancer (MBC) who had received ≥2 prior HER2-directed regimens for MBC. Methods: Patients were randomized 1:1 to N (240 mg qd po) + C (750 mg/m² bid po) or L (1250 mg qd po) + C (1000 mg/m² bid po). Co-primary endpoints were centrally assessed progression-free survival (PFS) and overall survival (OS). Secondary endpoints were investigator-assessed PFS; objective response rate (ORR); duration of response (DoR); clinical benefit rate (CBR); time to intervention for symptomatic metastatic central nervous system (CNS) disease; safety; and patient-reported health outcomes. Results: 621 patients were randomized (307 to N+C; 314 to L+C). The risk of disease progression or death was reduced by 24% with N+C vs L+C (HR = 0.76; 95% CI 0.63–0.93; p = 0.006); 6- and 12-month PFS rates were 47.2% vs 37.8% and 28.8% vs 14.8% for N+C vs L+C, respectively. OS rates at 6 and 12 months were 90.2% vs 87.5% and 72.5% vs 66.7% for N+C vs L+C, respectively (HR = 0.88; 95% CI 0.72–1.07; p = 0.2086). ORR in patients with measurable disease at screening was improved with N+C vs L+C (32.8% vs 26.7%; p = 0.1201), as was CBR (44.5% vs 35.6%; p = 0.0328) and DoR (HR = 0.50; 95% CI 0.33–0.74; p = 0.0004). Time to intervention for symptomatic CNS disease (overall cumulative incidence 22.8% vs 29.2%; p = 0.043) was delayed with N+C vs L+C. Treatment-emergent adverse events (TEAEs) were similar between arms, but there was a higher rate of grade 3 diarrhea with N+C vs L+C (24.4% vs 12.5%). TEAEs leading to neratinib/lapatinib discontinuation were lower with neratinib (10.9%) than with lapatinib (14.5%). Conclusions: N+C significantly improved PFS with a trend towards improved OS vs L+C. N+C also resulted in a delayed time to intervention for symptomatic CNS disease. Tolerability was similar between the two arms, with no new safety signals observed. Clinical trial information: NCT01808573