BACKGROUND: There is an increasing literature on learning organisations as a way of fostering communication, teamwork, collaboration and collective learning, thereby promoting quality improvement and enhancing patient safety. An increasing number of instruments are being developed in an attempt to measure learning organisation characteristics. However, the majority of these tools are created for a business setting, have not been scientifically tested and have not been applied in healthcare. OBJECTIVE: To evaluate elements of the validity and reliability of an instrument (ie, learning practice inventory (LPI)) for diagnosing learning practice characteristics in primary healthcare. METHOD: Content validity was evaluated using a modified nominal group technique and a content validity rating scale. Construct validity and reliability evaluation was undertaken with 10 staff members from 10 general practices in the west of Scotland. Staff completed the inventory twice, 4-6 weeks apart. Applying generalisability theory, a variance component analysis was performed. RESULTS: The main findings present evidence that the inventory has acceptable reliability and content validity. The results also demonstrate that the inventory can reflect the consistent and uniquely different perspectives of particular designations of staff within a practice. It is possible to compare practices' overall learning environments and to identify specific areas of practice strength as well as areas for development. CONCLUSION: This study demonstrates the psychometric properties of a learning practice diagnostic inventory. It highlights the consistently different perspectives that individual staff groups have on the function of their practice, suggesting that the success of quality improvement initiatives may be compromised without the involvement and true engagement of each staff group.