The comprehensive database of load tests on closed-ended piles in sand has been re-examined to study the relationship between CPT resistance, qc, and ultimate base capacity, qb. The aim is to establish the origin of low reported values of qb/qc which contrast with continuum models that suggest qb = qc for steady deep penetration. Partial embedment of the pile tip into a hard layer underlying weak material has been accounted for by weighting qc. Partial mobilisation has been accounted for by defining failure according to a plunging criterion. When these two mechanisms are considered, the resulting values of qb/qc have a mean value of 0·90 and show no trend with pile diameter. The remaining slight underprediction of the ‘continuum’ model (qb = qc) could be attributed to the underestimation of plunging load in pile tests for which steady penetration is not reached. This exercise makes two contributions: first, it is suggested that any reduction of qc when estimating the end bearing capacity of closed-ended piles in sand should be linked to partial embedment and partial mobilisation, rather than absolute diameter; second, the dearth of high-quality pile load test data in the public domain is highlighted.