CTLA-4 encodes cytotoxic T lymphocyte-associated antigen-4, a cell-surface molecule providing a negative signal for T-cell activation. CTLA-4 gene polymorphisms have been widely studied in connection with genetic susceptibility to various autoimmune diseases, but studies have led to contradictory results in different populations. This case-control study sought to investigate whether CTLA-4 CT60 and/or +49A/G polymorphisms were involved in the genetic predisposition to primary Sjögren syndrome (pSS). We analysed CTLA-4 CT60 and +49A/G polymorphisms in a first cohort of 142 patients with pSS (cohort 1) and 241 controls, all of Caucasian origin. A replication study was performed on a second cohort of 139 patients with pSS (cohort 2). In cohort 1, the CTLA-4 +49A/G*A allele was found on 73% of chromosomes in patients with pSS, compared with 66% in controls (p = 0.036; odds ratio (OR) 1.41, 95% confidence interval (CI) 1.02 to 1.95). No difference in CTLA-4 CT60 allelic or genotypic distribution was observed between patients (n = 142) and controls (n = 241). In the replication cohort, the CTLA-4 +49A/G*A allele was found on 62% of chromosomes in patients with pSS, compared with 66% in controls (p = 0.30; OR 0.85, 95% CI 0.63 to 1.16). Thus, the CTLA-4 +49A/G*A allele excess among patients from cohort 1 was counterbalanced by its under-representation in cohort 2. When the results from the patients in both cohorts were pooled (n = 281), there was no difference in CTLA-4 +49A/G allelic or genotypic distribution in comparison with controls. Our results demonstrate a lack of association between CTLA-4 CT60 or +49A/G polymorphisms and pSS. Premature conclusions might have been made if a replication study had not been performed. These results illustrate the importance of case-control studies performed on a large number of patients. In fact, sampling bias may account for some contradictory results previously reported for CTLA-4 association studies in autoimmune diseases.