Behavioural demand tests are informative tools for studying animal welfare, because they asses the motivation of an animal to obtain a given resource. Pekin ducks are a species on which behavioural demand testing has not been reported, despite the continued growth of the international duck meat industry and recognition that welfare issues exist for commercially farmed ducks. This study aimed to develop a behavioural demand technique specific to Pekin ducks that could have future application to assess their motivation for a variety of resources. Stage one of the study determined if ducks could learn an operant push-door task, and whether door design affected ducks’ willingness to complete the task. Sixteen female Pekin ducks (20 weeks old) were used, where they were presented with two door types (tall or short) in a cross-over experimental design. Twelve of the 16 birds successfully learnt the operant task. Non-parametric analyses (Wilcoxin signed rank test and Mann Whitney U test) were used to determine the effect of door type. Order of door presentation affected willingness to exit the behavioural demand unit (BDU). Birds presented with the short door first were less successful at exiting the BDU than when presented with the tall door (p < 0.0001) and took longer to exit from the short door (p = 0.03), whereas birds presented with the tall door first exited the short door with a similar amount of success and within a similar time period as the tall door. Stage two of the study validated the use of the chosen door type in the final design (tall door), by requiring ducks to exert increasing amounts of effort in order to return to their social group. Ten ducks that successfully completed the first stage were used in the second stage. Weights were progressively added to the push-door, so that ducks had to overcome between 10–100% of the average group bodyweight in order to return to their social group. Eight of the 10 birds pushed the maximum load presented, with the remaining two birds stopping at 80%. Increased physical effort occurred at 90 and 100% loads compared with the 10% load, with more pushes (p < 0.05) and longer exit latencies occurring (p < 0.05). We conclude that ducks are capable of learning operant tasks, and place value on having social contact. The successful design and validation of a behavioural demand method for Pekin ducks has been demonstrated, and should be suitable to investigate ducks’ motivation for a variety of resources.