ABSTRACTWe extend Burt, Burzynska, and Opper's cross-sectional network prediction of relative success among Chinese entrepreneurs by predicting which ventures are still active five years later. The cross-sectional analysis is corroborated in three ways (despite the vicissitudes of a national anti-corruption campaign): (1) Businesses run in 2012 by CEOs with a network rich in structural holes are more likely to be active five years later, in 2017. (2) Survival odds are improved if the large, open network around a CEO in 2012 was initially a supportive ‘cocoon’ closed network when the business was founded. (3) Both results are contingent on capturing theguanxities valuable early in the history of the business. The two network effects disappear when the network around a CEO is limited to his or her currently valued contacts. Beyond corroboration, we find that advantage is concentrated in ventures that began well and had become successful. Network advantage here does not compensate for weakness – it is a mechanism for cumulative advantage, amplifying the success of businesses already doing well.