Objective: To measure the intra- and inter-rater reliability of a biophotogrammetric assessment protocol for thoracoabdominal motion in preterm infants. Methods: This is an analytical cross-sectional study. Footage of 40 preterm infants was made in two views (lateral and anterior). The babies were placed in the supine position, with retroverted pelvis and semiflexed knees. Acrylic markers were positioned on surgical tape in eight predetermined anatomical points. We analyzed 4 variables in lateral view and 11 in anterior view (angular and linear) (ImageJ®), divided into two stages: 1. same frames - three blinded evaluators analyzed frames previously selected by the main researcher (inter-rater analysis 1), reviewing these same frames after 15 days (intra-rater analysis 1); 2. different frames - each evaluator selected the frames from the original video and repeated the protocol (inter-rater analysis 2), with a review after 15 days (intra-rater analysis 2). In stage 2, we tested the reliability of the entire process, from image selection to the analysis of variables. Data agreement and reproducibility were obtained by the intraclass correlation coefficient (ICC). Results: Agreement was high, particularly in angular variables (ICC 0.82 to 0.99). Linear variables ranged between very good and excellent in analysis 1 (same frames: ICC 0.64 to 0.99) and analysis 2 (different frames: ICC 0.44 to 0.89). Conclusions: The present study suggests that the proposed protocol for the thoracoabdominal motion analysis of preterm neonates has high reliability.