Inference with clustered survey data is important since multi-stage designs are often used to save time and costs. Variance estimation in the presence of multi-stage sampling is challenging. We consider pseudo-population bootstrap procedures. In the context of complete data, Chauvet (2007) proposed a p pseudo-population procedure and established its properties for smooth functions of means. In this presentation, we discuss pseudo-population bootstrap procedures with general clustered survey data. The proposed method can also be used in the presence of weight adjustment for unit nonresponse, imputation and weight trimming. Theoretical properties of bootstrap variance estimators will be discussed. Results from a limited simulation study, comparing the proposed approach with some existing ones, will be presented.