Abstract : Population structure influences importantly population growth rate and their ability to cope with environmental variations. Indeed, it determines the proportion of individuals who contribute to produce new generations as well as the intensity of competitive and cooperative social interactions. Using the common lizard (Zootoca vivipara, Jacquin 1787) as a model species, my work consists in the characterization of the relationship between population structure (in terms of cohorts, density and age and sex classes), life history and the dynamics of small isolated populations. To this aim, we combine experimental approaches in semi-natural conditions with modeling approaches using one-sex and two-sex age-structured population models. A critical review of the literature reveals that environmental variation can give rise to life-history differences among birth cohorts, or cohort effects, through its effects on individual quality and the intensity of intraspecific competition within and between cohorts. A first empirical study demonstrates that these three mechanisms generate life-history variation among cohorts, but also highlight the fact that direct effects of environmental variation on individual quality interact in a complex manner with indirect effects on the intensity of competition among individuals. During a second empirical study, we analyze the demographic processes and consequences of these competitive interactions. An elevation of density increases trophic competition and could induce hormonal responses to social stress, which reduce body growth and reproduction. A stochastic model demonstrates that this density dependence of life history interacts with the stochastic fluctuations of sex ratio to shape extinction dynamics of small populations of lizards. These predictions will be completed by the analysis of a third empirical study involving a joint manipulation of population density and sex ratio. Our results, integrated in the broad context of population ecology, highlight the importance of taking into account population structure and life-history heterogeneities in order to draw accurate predictions of the fate of small populations and to develop efficient management policies.