Approximately 1% of the population worldwide are affected by intellectual disability (ID), the vast majority of whom currently receive no molecular diagnosis. Previous studies indicate high levels of genetic heterogeneity, with estimates of over 2500 autosomal ID genes, with the majority being autosomal recessive (AR). Here, we combined microarray genotyping, homozygosity-by-descent (HBD) mapping, copy number variation (CNV) analysis, and whole exome sequencing (WES) to identify disease genes/mutations for 192 multiplex Pakistani and Iranian consanguineous families ascertained for non-syndromic ID. We identified definite or candidate mutations (or CNVs) in 51% of families, in 72 different genes, including 26 not previously reported for ARID. The new ARID genes include nine with loss-of-function mutations (ABI2, MAPK8, MPDZ, PIDD1, SLAIN1, TBC1D23, TRAPPC6B, UBA7, and USP44), and missense mutations include the first reports of variants in BDNF or TET1 associated with ID. The genes identified also showed overlap with de novo gene-sets for other neuropsychiatric disorders. Transcriptional studies show prominent expression in the prenatal brain. The high yield of AR mutations for ID signals that this approach has excellent clinical potential, will inform clinical diagnostics, including clinical whole exome and genome sequencing, for populations where consanguinity is common. As with other AR disorders, the relevance will also impact outbred populations.