Background: Children often play with more than one friend and there is no experimental evidence that has measured the effect of friendship groups on physical activity behavior. Therefore, the purpose of this study was to assess children's physical activity in a controlled setting during three, separate social conditions; alone, with a single friend and with a group (four) of friends. Materials and Methods: A total of 24 children between the ages of 6 to 10 years old (n = 12 boys, n = 12 girls) participated in three separate free-choice social conditions for 30-min. For each condition, children could select from a variety of physical and/or sedentary activities. Children's physical activity was the measured via accelerometry during each condition. Upon completion of each 30-min condition children were given the option to participate in an additional 10-min bonus period. Results: Accelerometer counts were not significantly (P < 0.05) different throughout each 30-min condition. However, during the optional, additional 10-min bonus period, children's physical activity significantly (P < 0.05) increased from the alone (713 ± 283 counts/min) to the friend (2,713 ± 339 counts/min) condition and then increased again from friend to the group (3,253 ± 147 counts/min) condition. Furthermore, a significantly greater (P < 0.001) proportion of children chose to participate in the 10-min bonus period during the friend (92%) and group (100%) conditions than the alone condition (33%). Conclusion: Playing with a single friend and a group of friends, relative to playing alone, is more motivating, and can increase physical activity when given the option for additional time. Additional research is warranted to experimentally assess how friendship groups can impact physical activity behavior during 60-min of free-choice activities.