We present a new 18-item parent-report measure of pretense development for 4- to 47-month-olds: the Early Pretending Survey (EPS). Study 1 (N = 231) demonstrated good internal reliability, and a strongly correlated 2-factor structure for 4- to 47-month-olds. Study 2 (N = 587) also showed good internal reliability, and found the two factors loaded onto a latent second-order Pretend Play factor. Study 3 (N = 84) found the EPS correlated with a researcher-led pretense experiment, however it no longer correlated when controlling for age. Additional analyses using subsamples of parents from Studies 2 and 3 showed good inter-observer reliability between parents (N = 29), and good longitudinal stability after 6 months (N = 196). Additional analyses combining Study 1–3 participants (N ≤ 902) found no item functioning differences across demographic variables, including child age, country, and socio-economic factors. Children’s EPS scores increased as they aged; girls had higher scores than boys; and children of younger parents had higher scores. Finally, we examined the age patterns of each EPS item, demonstrating when 25 %, 50 %, and 75 % of children were predicted to pass each pretense type. The EPS is useful for researchers to better understand how pretending relates to other areas of development (e.g., cognition, language). The EPS could also help parents, early years educators, and children’s media professionals decide which pretense types to use with children of different ages.