Objective: To develop a new observational code of parents playing with their infant in the first 18 months. Background: There is a lack of observational codes to analyse early interactions between parents and infants. We developed and tested this measure – the Parent Infant Play Observation code (PIPOc) – to be developmentally appropriate for infants, brief and easy to code after training, reliable and potentially suitable for clinical and research use. Method: The PIPOc was developed for use in a repeated measures trial of a new parenting programme. Mothers were filmed at home playing with their infant for 10 minutes with a six-month follow-up visit. The author also collected IT-HOME inventory data at the homes. A coding manual was developed to train an independent coder. Results: Six positive parenting behaviours were coded (Talk, Play, Touch, Move, Mind and Respond) with excellent to very good inter-rater reliability between the author and independent coder. Principal component analysis of the coded behaviours resulted in a three-component model termed -positive physical encouragement, sensitive parenting and verbal engegement components. A further analysis for the scores at follow-up resulted in the same component solution. Concurrent validity of the three components with subscale scores from the IT-HOME inventory subscales is reported. Conclusions: The PIPOc shows promising psychometric properties which are robust and reliable over six months. More extensive use of the PIPOc observation scores comparing intervention and control parents’ scores on the observation components will further test the code’s sensitivity to change over time.