This study tracked pre-service science teachers (PSTs) over 3 years as they completed required science courses (e.g., chemistry, biology and science and technology laboratory courses). The PSTs' use of argument, representations, and reasoning were examined (1) to determine the coherence within and between these and (2) examine the relationship between the use of these in relation to the degree to which the learning environment was generative. In this longitudinal, mixed methods study, a data-transformation variant of the convergent design was used. The study tracked nine PSTs and collected 522 laboratory reports over six semesters. The laboratory courses were classified into three levels as fully, partially and basic generative learning environments based on interviews and observations. The PSTs' arguments were analyzed considering the strength and connectedness of components; reasoning was examined using Walton's argument schemes; and representations were analyzed using multiple levels of representation in science. A Friedman test and multiple correlations were completed for statistical analysis. The results regarding the coherence within highlight that the utilization of argument, representation, and reasoning is strengthened when PSTs are in more generative environments, and the coherence between highlights a positive strong relationship between argument, representation, and reasoning, with the level of use being dependent upon the level of the generative environment. The importance of this study is that it demonstrates the need to be aware that merely labeling an environment as generative does not mean that benefits are the same. The more generative the learning environment, the greater the utilization of written reasoning, representation, and arguments.