PCV-2 within- and between-pen transmission was quantified by estimating the daily transmission rate $\beta$ and the basic reproduction ratio ($R_0$) using a stochastic SEIR (Susceptible, Exposed, Infectious, Removed) model fitted on experimental data. Within-pen transmission was quantified by using four groups of eight SPF (specific pathogen-free) pigs (four infected and four susceptible pigs having direct contact). Between-pen transmission was studied in two groups of 16 SPF pigs (eight infected and eight susceptible pigs having indirect contact (10 cm distance)). Pigs were monitored twice a week (blood samples) and were tested for PCV-2 antibodies (ELISA test) and viral genome load in sera (real-time PCR). Transmission parameters $\beta_{\rm within}$ and $\beta_{\rm between}$ were estimated using a maximum likelihood method and the duration of infectiousness, to compute $R_0$, was estimated with a parametric survival model. Different assumptions were made to determine the end of infectiousness (seroconversion, seroconversion and decline in viral genome load, permanent infectiousness). $R_{0[\rm within]}$ (8.9 (5.1-15.4)) was greater when the end of infectiousness was assumed to be related to both seroconversion and a decline of PCV-2 genome load in sera (average duration of infectiousness = 32 days) compared with only seroconversion as the indicator of recovery ($R_{0[\rm within]}$ = 5.5 (3.3-9.0)). Whatever the assumption, between-pen $R_0$ (0.58 (0.23-1.47)) was always significantly lower than within-pen $R_0$. Only $\beta_{\rm within}$ was sensitive to the assumption on end of infectiousness and decreased with increasing duration of infectiousness. These results showed that PCV-2 transmission is influenced by contact structure that appears worth being taken into account in an epidemic model.