This article examines kindergarten children's (5–6 years old) engagement in scientific practices, with a focus on generating and using evidence to support claims, during a 5‐month project about snails. The research questions are as follows: (1) what meanings do kindergarteners construct for what constitutes evidence? How are those meanings reflected in the development of data into evidence? (2) Which ways of gathering empirical evidence are jointly constructed by children and teacher during the project? (3) How do children use evidence to revise their understandings? The participants are one class of Early Childhood Education children (N = 25) and their teacher. They were engaged in a project about snails, involving pursuing their own questions, carrying out experiments and purposeful observations, collecting data and drawing conclusions, under the guidance of the teacher. The results show that children developed meanings of a certain level of sophistication about evidence, that they distinguished between empirical evidence from planned experiments and from prolonged observation, which we call purposeful, and that they combined different types of evidence in the revision of their ideas about snails. We identified two levels in the development of data into evidence—closer to descriptive statements and evaluative judgments. We suggest that purposeful observation, which has a clear focus, is guided by the teacher and explicitly discussed, has affordances in early childhood science. For instance, 30 out of 57 evidence statements relate to purposeful observation. Promoting purposeful observation as a source of evidence at this age may allow studying processes both for children (biology processes) and for researchers (learning processes). The results would support Metz's (2011) contention about the relevance of instructional opportunities over developmental constraints. © 2015 Wiley Periodicals, Inc. J Res Sci Teach 53: 1232–1258, 2016 Examínase a participación do alumnado de educación infantil (5–6 anos) nas prácticas científicas, en concreto en xerar e usar probas para sustentar conclusións, durante un proxecto de cinco meses sobre caracois. As preguntas de investigación son: (1) Que significados constrúen os nenos e nenas para o que constitúen probas? Como se reflicten estes significados no desenvolvemento de datos en probas? (2) Que formas de obter probas empíricas son construídas conxuntamente por nenos e mestra durante o proxecto? e (3) Como usan os nenos e nenas as probas para revisar o seu coñecemento? Os participantes son unha clase de terceiro curso de Educación Infantil (N = 25) e a súa mestra. Levaron a cabo un proxecto sobre caracois, procurando respostas ás suas propias preguntas, realizando experimentos e observacións cun propósito, recollendo datos e extraendo conclusións, guiados pola mestra. Os resultados mostran que desenvolveron significados de certa sofisticación sobre as probas, distinguindo entre probas procedentes de experimentos planificados e da observación prolongada, que denominamos cun propósito; e que combinaron diferentes tipos de probas na revisión das súas ideas sobre os caracois. Identificamos dous niveis na transformación de datos en probas, enunciados cercanos a descricións e xuízos avaliativos. Suxerimos que a observación cun propósito, caracterizada por ter un obxectivo definido, estar guiada pola mestra e ser discutida explicitamente, ten potencial no ensino das ciencias en educación infantil e primaria. Por exemplo, 30 dos 57 enunciados sobre probas relaciónanse coa observación cun propósito. Promover a observación cun propósito como fonte de probas nestas idades pode permitir estudar procesos, tanto polos nenos (procesos biolóxicos) como polas investigadoras (procesos de aprendizaxe). Os resultados sutentan a perspectiva de Metz (2011), respecto da relevancia da instrución sobre limitacións debidas ao desenvolvemento.