Testing of programs is important to increase the reliability of programs. Coverage is a ratio of the number of worked test events to all test events, and it is used as a metric of testing sufficiency and reliability. The test events are defined by a testing criterion. Some testing criteria are proposed for evaluating testing sufficiency of sequential programs. However, the criteria are inadequate for concurrent programs. New testing criteria must be introduced for concurrent programs. The paper proposes new testing criteria, ordered sequence criteria (OSC) for concurrent programs. OSC are concerned with interprocess communication and synchronization. An OSCk selects k-length sequences of statements related to communication or synchronization. The sequences should be executed at least once in testing. OSCk presents various levels of testing according to values of k. The OSC2 is reliable for a program which is correct or which includes communication errors. A prototype is implemented for coverage measuring based on OSC2.