The aim of this study was to analyze the relationships between four personality traits (calmness, trainability, dog sociability and boldness) of dogs (Canis familiaris) and dog and owner demographics on a large sample size with 14,004 individuals. German speaking dog owners could characterize their dog by filling out a form on the Internet. There were five demographic variables for dogs and nine for owners. Two statistical methods were used for investigating the associations between personality and demographic traits: the more traditional general linear methods and regression trees that are ideal for analyzing non-linear relationships in the structure of the data. The results showed that calmness is influenced primarily by the dog's age, the neutered status, the number of different types of professional training courses (e.g. obedience, agility) the dog had experienced and the age of acquisition. The least calm dogs were less than 2.5 years old, neutered and acquired after the first 12 weeks of age, while the calmest dogs were older than 6.9 years. Trainability was affected primarily by the training experiences, the dog's age, and the purpose of keeping the dog. The least trainable dogs had not received professional training at all and were older than 3 years. The most trainable dogs were those who participated in three or more types of professional training. Sociability toward conspecifics was mainly determined by the age, sex, training experience and time spent together. The least sociable dogs were older than 4.8 years and the owners spent less than 3h with the dog daily. The most sociable dogs were less than 1.5 years old. Males were less sociable toward their conspecifics than females. Boldness was affected by the sex and age of the dog and the age of acquisition. The least bold were females acquired after the age of 1 year or bred by the owner. The boldest dogs were males, acquired before the age of 12 weeks, and were younger than 2 years old. Other variables, including the owner's gender, age, education, previous experience with dogs, the number of people and dogs in the household, and purpose of keeping the dogs had minor, but detectable effects. The results suggest that on-line questionnaires are a very effective means for collecting data about dog behavior, especially if owners are motivated by instant feedback. However, note that the characteristics of dogs in the present study were reported by the owners, and the associations with the traits do not necessarily represent a causal relationship.