Parent birds are expected to show anti-predator responses when predators are in the vicinity of their fledglings and to modify their response in relation to perceived risk posed by the predator. We used the superb fairy-wren (Malurus cyaneus) to experimentally test predictions of the risk-based alarm call hypothesis, whereby alarm vocalisation response is a proxy for predator risk (type, distance). Our results showed that birds modified their response to three factors: predator type (snake, fox, stationary and gliding sparrowhawks), predator distance (close, distant) and fledgling presence. We found evidence of post-fledging parental care in response to the fox, which was significantly higher when fledglings were present irrespective of predator distance. However, fledgling presence was not related to alarm vocalisations to the snake or the sparrowhawks (only distance predicted vocalisation response). A comparison of the different types of vocalisations (terrestrial call, aerial call, alarm song) showed that alarm vocalisations were significantly related to predator type. Fledgling presence also affected the frequency of parental terrestrial alarm calls. We conclude that anti-predator response is a dynamic process that reflects offspring presence and perceived predation risk, with implications for understanding vocal communication in birds.