This research is a kind of research and development that aims to describe the process and result of development problem-based learning device on social arithmetic, describe the effectiveness of problem-based learning on social arithmetic, compare the mathematical disposition of students who follow problem based learning and mathematical disposition of students who follow the conventional learning. The procedures refer to the Plomp development model that includes: preliminery investigation; design; realization; test, evaluation, revision; implementation. Descriptive analysis and inferential statistical analysis are used to analyze the research data. The findings were as follows: the validation of the learning device has a good category value; each item test result are valid, reliable, sensitive; observation result of learning meet good criteria; the student activity are effective; the student response are positive, mastery learning are achieved; and the results of statistical tests show that two regression models are linear, not equal and parallel. Based on the analysis, it can be concluded that the resulting learning device meet the criteria valid, practical, effective; problem based learning is effective to teach social arithmetic and the mathematical disposition of students who follow problem based learning are better than the mathematical disposition of students who follow conventional learning.