We assessed the efficacy and safety of capecitabine across treatment lines, and the impact of patient and disease characteristics on outcomes using data from phase II/III trials. Individual patient data were pooled from seven Roche/Genentech-led trials conducted from 1996 to 2008 where single-agent capecitabine was the test or control regimen for metastatic breast cancer (MBC). Data were analyzed from 805 patients: 268 in the first-line metastatic setting and 537 in the second-line or later setting. Baseline characteristics were balanced across treatment lines. Patients receiving second-line or later versus first-line capecitabine had lower objective response rates (ORR: 19.0 vs. 25.0 %, respectively, odds ratio 0.70; 95 % CI: 0.5-1.0) and significantly shorter progression-free survival (PFS: median 112.0 days [3.7 months] vs. 150.0 days [4.9 months]; p < 0.0001) and overall survival (OS: median 396.0 days [13.0 months] vs. 666.0 days [21.9 months]; p < 0.0001). In multivariate analysis by backward elimination, significantly improved ORR (p = 0.0036), PFS (p < 0.0001) and OS (p < 0.0001) with capecitabine were demonstrated in patients with estrogen receptor (ER) and/or progesterone receptor (PgR)-positive versus both ER and PgR-negative tumors. Hand-foot syndrome (HFS) was the most common adverse event (AE) in 63 % of patients. Overall, 7 % of patients discontinued and two patients (<1 %) died from treatment-related AEs. Significantly improved survival was observed in patients developing capecitabine-related HFS (p < 0.0001 PFS/OS) or diarrhea (p = 0.004 OS; p = 0.0045 PFS) versus patients without these events. In this pooled analysis of individual patient data, first-line capecitabine was associated with improved ORR, PFS, and OS versus second or later lines. Multivariate analyses identified greater ORR, PFS, and OS with capecitabine in patients with ER and/or PgR-positive versus ER/PgR-negative tumors. Safety was in-line with previous phase III trials in MBC.