CONTEXT: Ingestion of toxic liquids is common, and the volume ingested is often important for clinical decision-making. However, the accuracy and interpretation of volume estimates in the context of toxicological exposures is poorly characterised in adult practice. OBJECTIVE: To inform the interpretation of volume estimates when expressed in forms commonly encountered in toxicological practice: (1) semi-quantitative volume descriptors, such as 'mouthfuls'; (2) quantitative self-estimates of ingestion volume, for example, millilitres; and (3) estimates of residual volume in containers. METHODS: In the first part of the study, 50 members of the public ingested water in response to requests to take a 'small mouthful', 'large gulp' and 'five mouthfuls'. They estimated the amount ingested, and actual volumes were measured. In part 2, 15 members of the public and 15 healthcare professionals estimated the volumes contained in 12 opaque and transparent bottles. RESULTS: The mean age of participants in part 1 was 37 years, and in part 2 it was 34 years. The mean volume (95% prediction interval) of a 'small mouthful' was 43 (3-137) mL; 'large gulp', 77 (20-168) mL; and 'five mouthfuls', 157 (25-375) mL. The mean error (95% limits of agreement) for self-estimates of ingestion volume was an underestimate of - 52% (- 90% to + 124%). Volume contained in bottles was underestimated by - 5% (- 38% to + 27%). This varied according to the container type (mean difference: opaque, - 10%; transparent, - 1%; P < 0.01) and participant type (members of the public, - 8%; healthcare professionals, - 3%; P = 0.02). CONCLUSIONS: Volume estimates derived from semi-quantitative descriptors are not a reliable basis for clinical decision-making. Self-estimates provided in a quantitative form are inaccurate and prone to underestimation. Estimates of residual volume in containers should be regarded as suspect if the container is opaque. Where clinical decisions hinge on the volume ingested, efforts should be made to quantify this using measurement.