The current research aims at constructing a developmentally sensitive mindreading scale (i.e., a battery of tasks measuring different aspects of mindreading ability in children from 1 to 3.5 years of age). Over 300 Polish children were tested at six-month intervals with 48 different tasks designed to measure mindreading ability (for a total of six measurement occasions, T1-T6; NT1 = 357; MageT1 = 52,30 weeks; SD = 1,73; 201 (56%) boys, 156 (44%) girls). Using latent trait modelling, a latent mindreading factor was established. Support was found for satisfactory reliability of the scale that ultimately consisted of 31 tasks. The tasks were age-matched, as the information curves for latent factors for each of the six age ranges and the reliability of the composite constructs were mostly acceptable. The validity was also confirmed using correlational analyses, and the continuous development of mindreading was observed over the tested age ranges: from joint attention, through using gestures in communication and visual perspective taking, to false-belief understanding. The proposed scale is suitable for use in the investigation of the general pattern of the development of, and individual differences in, mindreading ability.