This article examines some of the considerations that should be made to successfully generate meaningful TTCN [1] test cases from SDL [2] designs. A number of criteria for the test case generation are identified. These criteria depend both on how state space exploration is performed, as well as on a set of other conditions. A number of state space exploration algorithms and their impact on the generated set of test cases are described. Observer processes and how they can be defined to improve test case generation are also examined. The chosen examples that are examined are based on industrial implementations, using the Telelogic Tau toolset (http://www.telelogic.com) that handles all the languages and methods described below.