The use of formal observation in primary mathematics classrooms is supported in the literature as a viable method of determining effective teaching strategies and appropriate tasks for inclusion in the early years of mathematics learning. The twofold aim of this study was to (a) investigate predictive relationships between primary mathematics classroom observational data and student achievement data, and (b) to examine the impact of providing periodic classroom observational data feedback to teachers using a Relational-Feedback-Intervention (RFI) Database Model. This observational research effort focused on an empirical examination of student engagement levels in time spent on specific learning activities observed in primary mathematics classrooms as predictors of student competency outcomes in mathematics. Data were collected from more than 2,000 primary classroom observations in 17 primary schools during 2009-2011 and from standardised end-of-year tests for mathematics achievement. Results revealed predictive relationships among several types of teaching and learning tasks with student achievement. Specifically, the use of mathematics concepts, technology and hands-on materials in primary mathematics classrooms was found to produce substantive predictors of increased student mathematics achievement. Additional findings supported the use of periodic classroom observation data reporting as a positive influence on teachers' decisions in determining instructional tasks for inclusion in primary mathematics classrooms. Study results indicate classroom observational research involving a RFI Database Model is a productive tool for improving teaching and learning in primary mathematics classrooms.