Purpose: This study investigated the auditory comprehension of Japanese sentences including relative clauses (RCs) by 52 Japanese-speaking children with typical development (TD) and 16 children with specific language impairment (SLI). Method: A picture-pointing task measured RC and main clause (MC) comprehension for object and subject relatives in sentences with varying lengths. The accuracy of RC only, MC only, and combined comprehension (RC and MC) was analyzed with linear mixed-effects (logistic) models to examine accuracy and the effects of the deep gap-filler distance (structural complexity), the surface linear distance, and sentence length for the two participant groups. Six language tests and two working memory tests were administered to determine language and working memory status. A factor analysis grouped them into two factors that were examined as predictors of comprehension performance. Results: Initial models indicated that children with SLI performed more poorly than their peers with TD. Working memory abilities predicted the comprehension of subject and object relative sentences, but language abilities other than a general measure of sentence comprehension were only limited predictors. In a series of pairwise comparisons, children with SLI did not exhibit comprehension facilitation for RC sentences with lower structural complexity, nor did they exhibit comprehension facilitation when the linear distance between the gap and the filler was shorter, perhaps because of their overall lower performance. Children with TD made more errors on sentences with higher structural complexity but did not exhibit any linear distance effects. Neither group exhibited effects of noun phrase or sentence length. Conclusions: Japanese-speaking children with TD exhibited negative comprehension effects of increased deep structure complexity, but not of linear surface distance or sentence noun phrase length. The children with SLI face challenges in comprehending both subject and object relative sentences compared to their typically developing peers. Their poorer working memory abilities are the primary factor that constrains their performance.