Familial clustering of cancer is expected to occur at practically all anatomical sites. However, few studies have had sufficient size to investigate different sites simultaneously and with adjustment for confounders. We evaluated familial clustering in the Netherlands Cohort Study in which 120,852 men and women, aged 55-69 years in 1986 were followed up for 13.3 years. 14,025 Probands, 6,629 parents and 4,271 siblings were diagnosed with cancer. Relative Risks (RR) of cancer in first degree family members were calculated by using multivariable Cox regression analyses. We also calculated false-positive reporting probabilities. Significant concordant familial clustering was observed for stomach (RRfather = 1.89, RRparent = 1.66, RR sister = 3.33, RRsibling = 2.38, RR1st degree = 1.69), colon/rectum (RRfather = 1.82, RRmother = 1.83, RRparent = 1.88, RR1st degree = 1.56), lung (RR brother = 1.50) and breast cancer (RRmother = 1.65, RRsister = 1.72, RR1st degree = 1.72) With low false-positive reporting probabilities. Significant discordant familial clustering has been observed for combinations of pancreas-colon/rectum (RR mother = 2.42, RRparent = 1.89, RR1st degree = 1.73), larynx-lung (RRfather = 3.35, RRparent = 2.84, RR 1st degree = 2.30), lung-oesophagus (RRsibling = 3.49), breast-bladder (RRfather = 2.79, RRparent = 2.61), endometrium-stomach (RRmother = 2.32), ovarium-oesophagus (RR 1stdegree = 4.19), prostate-colon/rectum (RRparent = 1.46) and bladder-larynx/pharynx (RRfather = 2.49) cancer, although false-positive reporting probabilities were higher for these associations. Familial clustering of cancer occurs at most sites but is generally modest. Some observed discordant familial clustering is surprising but should be interpreted with caution as their prior probability is low. © 2008 Wiley-Liss, Inc.