The purpose of this study is to assess the learning effect of a multilingual web-based argumentative writing instruction model called the Ensayo Científico Multilingüe (ECM, Multilingual Scientific Essay) adapting the didactic model called Genre-based Writing Instruction (GBWI) in an experiment conducted over three months. For this purpose, a quasi-experimental research model was applied to 150 students in the experimental group and 150 in the control group, with two measurements, pre and post-test, for three dependent variables: (a) writing metacognition and its dimensions; (b) written argumentative self-efficacy; and (c) rhetorical moves and steps of an argumentative essay. The latter variable was measured by the content analysis method. Variables (a) and (b) were both measured with instruments validated in a population of 518 university students using structural equations. The findings demonstrate the positive effect of the ECM, which combines WBWI and GBWI in argumentative written learning in the students’ mother tongue in all variables measured, applying statistics such as the Shapiro–Wilk statistic, parametric contrast, and the Wilcoxon signed-rank test. In relation to the findings, with respect to the evaluated variables, it was discovered, specifically, that the rhetorical steps in which the students showed a significant improvement were innovations, quotes/research, definitions of concepts, refutations, definitive reasons, and bibliographical references. Likewise, the rhetorical steps that did not present significant differences following the application of the ECM were discovered, and they were: reason summary, formulation of premise, and reasons for. Furthermore, it can be stated that for the ECM there was an increase, above all, in awareness of the following metacognitive dimensions: (a) writing self-regulation; (b) writing planning; and (c) writing revision, as well as argumentative self-efficacy. The novelties of this research with respect to the precedents reside in that it offers valid and concrete results on the effect of a multilingual web design integrated into a well-defined didactic model of argumentative writing on writing metacognition and its dimensions, argumentative structuring and its rhetorical steps, and argumentative self-efficacy. The related studies consider only some of these variables, but not all of them together or their complexity. These results have allowed us to establish specific didactic–technological proposals for improving the ECM that are transferable to didactic designs to guide written argumentation at higher academic levels using multilingual web technologies and integrating the metacognitive, behavioral, and motivational dimensions of writing.