Positive animal emotion (affect) is a key component of good animal welfare  and plays an important role in stress-coping and resilience . Methods for reliably inducing and measuring positive affect are critical, but both have been limited in availability. In rats, one promising way of inducing positive affective states is by human-simulated rough and tumble play or ‘tickling’ [3,4]. However, in humans tickling induces both pleasure and displeasure, and neither an established non-verbal indicator of positive affect, the Duchenne smile, nor laughter detects this variation [5,6]. Rats also show individual differences in response to tickling , and this variation needs to be readily quantified if we are to ensure that tickling is only implemented where it generates positive affect. Here, we use a validated and objective measure of affective valence, the affective bias test , to show that 50 kHz ultrasonic vocalizations provide a quantifiable and graded measure of positive affect that accurately reflects the positive state induced by this human–rat interaction.