An instrument to assess primary school science content knowledge in pre-service primary schoolteachers was elaborated and validated. A multiple-choice test with 50 items was initially proposed. Four content areas of science from the curriculum of Primary Education were considered: natural environment and its conservation, biological diversity, health education and the human body, matter and energy. A sample of 83 pre-service teachers from a Spanish university participated in the statistical validation process. External consistency (Test-Retest correlation), internal consistency (Kurder-Richardson's 20th formula), item difficulty and item discrimination indices were analysed as well as distractor selection. Some adjustments were made and a second shorter 30-item version was proposed. This new instrument was validated with four different samples amounted for a total of 756 Spanish pre-service primary teachers. Distributions of difficulty and discrimination indexes were analysed. Consistency among samples was very good. Remarkable results about participants' poor science content mastery and misconceptions were shown. Implications of these findings are discussed.