Most Spanish learner corpora are composed of texts written by students who complete one or more tasks once, at a specific moment of their linguistic development. Thus, in studies that look into the lexical or grammatical characteristics of learners at different levels of proficiency, analyses are based on comparisons between different groups of students at each level. This is a source of variability, as it is impossible to determine whether observed differences are really due to the proficiency level of the learners or to the individual differences between the participants who completed the tasks in each level. The alternative is to obtain longitudinal data, which makes it possible to observe the actual learning trajectory of individual learners across 375time. However, collecting this type of data poses additional burdens, given that few learners would accept to participate in the data collection process at several time points, with weeks, months or years of interval between them. This chapter aims to (1) present three Spanish learner corpora that do include longitudinal data: LANGSNAP, Aprescrilov II y COWS-L2H, (2) analyze the methodological differences between them, as well as the advantages and disadvantages of each approach and (3) illustrate the complementarity of longitudinal and cross-sectional data by studying the developmental patterns in lexical diversity observed through both types of data in COWS-L2H.