The severity of liquefaction-induced ejecta manifestation for the 2010–2011 Canterbury earthquakes was overestimated or underestimated using simplified liquefaction ground-failure indices at several sites in Christchurch. Nonlinear effective stress analyses of two representative level ground sites are performed to investigate the causes of these misestimations. One site has a thick, clean sand deposit, and the other site has a highly stratified silty soil deposit. The excess hydraulic head profile with depth that develops during and after earthquake shaking determines the potential of upward seepage-induced artesian flow that produces sediment ejecta. The thick, clean sand site can develop high-gradient upward seepage that is sustained after strong shaking ends to produce severe ejecta. The stratified silty soil site develops high pore water pressures in isolated soil layers, but the amount and rate of upward seepage are insufficient to produce ejecta. A proposed ejecta potential index (EPI) captures key aspects of the hydraulic mechanisms of liquefaction manifestation. EPI estimates the severity of sediment ejecta by tracking the duration in which the excess hydraulic head exceeds the hydraulic head required for artesian flow. The EPI values computed from the simulations of the two sites capture the observed trends of liquefaction manifestations during the Canterbury earthquakes.