This paper considers victim heterogeneity in harm levels in a bilateral-care model. In practice, resources are expended on the verification of the accurate magnitude of damages suffered. We show a suficient condition for the possibility to accurately deduce the harm level from the observable care choice without spending on verification or other facilitating means. For cases in which this condition does not hold, this paper sets out a simple screening mechanism that induces victims to reveal their type truthfully and induces optimal care in equilibrium without verication costs.