In Australia, one of the key findings of the Teacher Education Ministerial Advisory Group was that not all graduating pre-service teachers possess adequate pedagogical content knowledge (PCK) to teach effectively. The concern is that higher education providers working with pre-service teachers are using pedagogical practices and assessments which are not informed by research. Due to its multifaceted nature, PCK is difficult to measure. In the case of mathematics PCK, some researchers have used multiple-choice questions while others use constructed-response formats depending on the specific attribute of PCK being measured. In either case, researchers need to use instruments that are appropriate to each task. This paper reports on the construction and reliability of a survey instrument used in a larger study to measure pre-service teachers' mathematics PCK. Using four stages of survey development, in addition to inter-rater agreement measures and Rasch modelling techniques, data analyzes found high levels of validity and reliability for the Mathematics Pedagogical Content Knowledge Instrument (MPCKI).