Background: The implicit learning deficit hypothesis claims that impaired implicit learning underlies deficits in social-communicative abilities associated with developmental language disorder (DLD). However, previous research testing this hypothesis revealed inconsistent results and largely used process-impure sequential learning tasks. Aims: This study further tested the hypothesis using a novel process-pure implicit associative learning task. Methods and Procedures: The performance of 9- to 13-year-old children with (N = 60) and without DLD (typically developing, TD, N = 52) on a contingency learning task (CLT) was compared. The task entailed the incidental learning of the contingency between simultaneously presented figure-colour combinations. Also, the association of CLT performance with three aspects of social-communicative abilities was assessed: facial emotion recognition ability, social responsiveness and language abilities. Outcomes and Results: Compared to the TD group, the DLD group performed equally on the CLT but showed worse performance on the measures of emotion recognition and social abilities. In neither group was CLT performance significantly related to any of the three social-communicative abilities. Conclusions and Implications: These results do not support the implicit learning deficit hypothesis. The demonstrated intact implicit learning ability suggests the potential of using interventions to improve social-communicative abilities in children with DLD that are based on incidental or implicit learning rather than on intentional or explicit learning.