Background Given the significance of early language for later academic achievement, language development can be an important outcome measure in evaluation of early childhood intervention programmes. Language development may be challenging to assess in settings where trained personnel are hard to find; therefore, maternal report can be a reliable alternative measure. Aim The aim of this study was to adapt a maternal report measure, the MacArthur Communicative Development Inventory II‐Short Form, in the Sindhi language and to examine its psychometric properties as a measure of language outcome in 2‐year‐old children in rural Pakistan. Methods A three‐phase process was followed to produce a Sindhi adaptation of the tool comprising 100 words. The first phase was a review of the original 258 words through a focus group discussion with a team familiar with the context; the second phase was testing 258 words to compile a 100‐word list of easy, moderate and difficult words. The third phase was a pilot of the 100‐word list followed by administration in a larger intervention study population. Results On administration with 1381 children, concurrent validity with the Bayley Scale of Infant Development III using the Pearson's correlation test showed a moderate association for comprehension (r = 0.45, p = 0.01, n = 1381) and good association for expression (r = 0.51, p = 0.01, n = 1381). Internal consistency was high with alphas of 0.98 for comprehension items and 0.96 for expression items. Conclusion The authors conclude that the Sindhi adaptation of the MacArthur Communicative Development Inventory followed a feasible and rigorous methodology to create a reliable and sensitive tool to assess young children's language development for use in a child assessment battery for early childhood health, nutrition and development studies.