This study aimed to identify characteristics of dogs to be used by the New Zealand Police Section for their dog breeding and selection programs. Only one scientific paper has been published on selecting dogs specifically for police dog work. Currently, no statistical analysis has been conducted on any of the New Zealand Police Dog Section's data. A questionnaire on all aspects of police dog work was sent to the 120 operational police dog handlers working in New Zealand. The majority of handlers rated their dog high for a number of traits and areas of police dog work but their ideal dog rated very high for the same traits. The handlers ranked from highest to lowest the traits 'prey drive', 'trainability', 'activity', 'obedience', 'playfulness', 'independence' and 'aggressiveness'. This gives an indication of the relative emphasis that should be given to the traits in a selection program. Improved stud selection, better monitored foster homes, more consistency between regions and the training centre and having more dogs for selection are improvements that can be made. Annual reports from dog trials for the years 1997 to 2000 were analysed by ANOVA to enable the calculation of repeatabilities for each activity. The activities 'heel free', 'retrieve', 'down stay', 'sendaway', 'recall and redirection', 'distance control', 'speak on command', 'track', 'article search', 'passive attack', 'chase and recall', 'chase and attack', and 'control' were measured in all four annual reports. The activity 'search and escort' was measured in 1999 and 2000. The highest repeatability (0.48) was for 'speak on command' and the lowest repeatability (0.03) was 'track'. There were insufficient data to enable the estimation of heritability values. During the annual trials each activity should be separated into handler performance and dog performance to give an indication of the performance of the dog alone. If the traits essential for each trial activity were identified and measured when the activity was being tested then a repeatability study on the trait alone could be conducted. This report identifies several areas where changes in trait definition and the collection of information could be used to improve the efficiency of the police dog breeding program.