RATIONALE: Multiple observational coding systems have been developed and validated to assess parent-child interactions during painful procedures. Most of these coding systems are neither theory-based nor do they well represent parent nonverbal behaviours. AIMS: Develop the Parent Caring Response Scoring System (P-CaReSS) based on Swanson’s Theory of Caring and test its psychometric properties in children in cancer port starts. METHODS: A hybrid approach of inductive and deductive coding was used to formulate the preliminary observational codes for the P-CaReSS. Twenty-nine children, each with one video-recording of port start available, were selected from the parent study (R01CA138981) to refine the P-CaReSS, train coders and test inter-rater reliability. Videos of another 43 children were used to evaluate the construct validity of P-CaReSS. Per cent agreement and Cohen’s kappa were used to present the inter-rater reliability. Spearman rank-order correlations were used to report the construct validity. RESULTS: The 18-item P-CaReSS includes three types of parent behaviours: verbal, nonverbal and emotional behaviours. These parent interaction behaviours comprise five caring domains – knowing, being with, doing for, enabling, and maintaining belief – and one noncaring domain. On average the per cent agreement was 0.82 for the P-CaReSS overall, with average per cent agreements above 0.80 for both verbal and nonverbal behaviours. Kappa coefficient was 0.81 for the emotional behaviour. The behavioural codes in the P-CaReSS showed significant correlations with independent ratings of parent distress, child distress and child cooperation. CONCLUSIONS: The P-CaReSS is a promising tool that can be used to evaluate parent verbal, nonverbal and emotional behaviours during cancer-related port starts. This observational tool can be used to guide the development of nursing interventions to help parents caring for their child during cancer procedures.