Glycosylation of cellular proteins has important impact on their stability and functional properties, and glycan structures strongly influence cell adhesion. Many enzymes are involved in glycoconjugate synthesis and degradation, but there is only limited information about their role in breast cancer progression. Therefore, we retrieved RNA expression data of 202 glycosylation genes generated by microarray analysis (Affymetrix HG-U133A) in a cohort of 194 mammary carcinomas with long-term follow-up information. After univariate and multivariate Cox regression analysis, genes with independent prognostic value were identified. These were further analysed by Kaplan-Meier analysis and log-rank tests, and their prognostic value was validated in a second cohort of 200 tumour samples from patients without systemic therapy. In our first cohort, we identified 24 genes with independent prognostic value, coding for sixteen anabolic and eight catabolic enzymes. Functionally, these genes are involved in all important glycosylation pathways, namely O-glycosylation, N-glycosylation, O-fucosylation, synthesis of glycosaminoglycans and glycolipids. Eighteen genes also showed prognostic significance in chemotherapy-treated patients. In the second cohort, six of the 24 relevant genes were of prognostic significance (FUT1, FUCA1, POFUT1, MAN1A1, RPN1 and DPM1), whereas a trend was observed for three additional probesets (GCNT4, ST3GAL6 and UGCG). In a stratified analysis of molecular subtypes combining both cohorts, great differences appeared suggesting a predominant role of N-glycosylation in luminal cancers and O-glycosylation in triple-negative ones. Correlations of gene expression with metastases of various localizations point to a role of glycan structures in organ-specific metastatic spread. Our results indicate that various glycosylation reactions influence progression and metastasis of breast cancer and might thus represent potential therapeutic targets.