The cost to companies of training first time users of a package is high, especially if the claims made by MacGregor (1993) are true. In a newspaper article reviewing the WordPerfect package, he suggests that 95% of the package's features will never be used by 98% of its users. As this wordprocessing package is one of the most widely used packages, there is no reason to believethat this might be any different for less frequently used packages. This less than optimal use of package capabilities, which was also found by Czaja (1986), may be a result of conceptual hurdles which are encountered by users and never surmounted. These may never have been surmounted as a consequence of not having been recognized and addressed in software design and training material for same. In an exploratory study which examined users interacting with one of two statistical package for the first time,users were found to experience various conceptual hurdles. Although the packages had different requirements for executing the simple data entry task assigned, the same conceptual hurdles were noted in each. These were evident from the analyses of the detailed keystroke level traces collected from each user. The packages were Merlint, a mixed command structure package, and a command driven version of Minitabt