Labile characters, like behaviors, are phenotypes that are expressed repeatedly in the life of an individual. These types of characters allow individuals to adjust their phenotype to various levels of environmental variation, and therefore play a key role in the evolutionary process. Labile phenotypes are distinct because of their multi-level nature; individuals can differ in their average phenotypic expression (causing among-individual variation), but they can also vary their phenotype in each expression (causing within-individual variation). In order to understand the role of labile characters in the evolutionary process it is necessary to acknowledge that variation at each level is caused by different processes. Variation at the among-individual level is caused by genetic or environmental differences having a permanent effect on an individual’s phenotype, whereas variation at the within-individual level is caused by an individual’s adjustment of its phenotype to a changing environment. The implications of these multi layered effects in the expression of labile characters have been acknowledged by different fields of evolutionary ecology, but major areas of evolutionary research do not fully incorporated this idea. The general aim of my thesis was to fully integrate this multi-level nature in the study of the adaptive causes and evolutionary consequences of variation in labile characters. My thesis is composed of five chapters: the first three are conceptual and methodological works aimed at integrating the multi-level nature of labile characters into already existing evolutionary frameworks. The last two chapters describe, as a worked example, how the different levels of variation and covariation between (labile) fertilization related traits affect the evolution of the alternative reproductive strategies in a wild passerine bird (the great tit). The first chapter is a conceptual work focusing on how to define and statistically characterize behavioral characters. We argue that behavioral characters can be studied using the “evolutionary character concept”. This framework was developed to study characters that only vary among individuals (i.e. “fixed characters”); therefore we extended this framework to include characters that also vary within-individuals. The second chapter of the thesis is a methodological work where we proposed a way to quantify multi-level variation in reaction norms, which allows the estimation of repeatability of plasticity. Behavioral ecologists have recently developed theory predicting the ecological conditions where repeatable vs. non-repeatable variation in phenotypic plasticity should evolve. However, there was no methodological framework to estimate repeatability of plasticity. Therefore, we proposed a study design and mixed effect model structure to estimate repeatability of plasticity. To help researchers use the proposed methodology, we developed an R simulation package to estimate bias, precision and accuracy for different sampling designs. The third chapter is an opinion paper that urges researchers to combine theory and methods developed in behavioral ecology and quantitative genetics to study phenotypic variation in a social context. Quantitative geneticists have developed a framework to study social evolution aimed at predicting the evolutionary response to selection of traits affected by the phenotypes of other individuals (the “social environment”). Phenotypes expressed in a social context, also called interactive phenotypes, exhibit a particular evolutionary dynamic because their environmental component is composed of genes and can thus evolve. Despite that fact that the effects of the social environment are commonly mediated by labile characters, this social evolution framework has not fully considered the multi-level nature of labile characters. Therefore, for chapter three we integrated the multi-level nature of labile characters in this social evolution framework. The final tw