SPINK1 has been described to be mutually exclusively expressed in prostate cancer (PCa), but its expression profiles and the probable roles in bone metastatic PCa have not been thoroughly explored. Total of 155 biopsy specimens from initially diagnosed bone metastatic PCa were obtained between 2009.1 and 2012.12. SPINK1 and ERG were detected by using immunohistochemical staining. Factors included age, ECOG score, clinical T stage, Gleason scores (GS), expression of SPINK1 and ERG, baseline PSA, baseline ALP, baseline HGB and PSA normalization, and the association of SPINK1 and ERG with clinical outcomes (CRPC-free survival and overall survival) were analyzed. Totally, SPINK1 and ERG were mutually independently expressed in the primary tissues of those patients, and their positivity were only 13.5% (21/155) and 10.9% (17/155), respectively. Positive expression of SPINK1 was completely detected in cases with primary Gleason score 4 or 5; on the contrary, the frequency of ERG was much lower. Correlative analysis only found that SPINK1 was linked with PSA response to androgen deprivation therapy (χ(2)  = 11.101, P = 0.001). Survival analysis showed that, ERG was not associated with clinical outcomes in all cases, especially in cases with higher GS (8-10) (n = 90); but SPINK1 was an independent prognostic factor which was associated with adverse CFS of patients with GS 8-10 (CFS: HR = 5.141, 95%CI: 1.108-25.552, P = 0.017). It is the first time to simultaneously detect SPINK1 and ERG expression in initially diagnosed bone metastatic PCa. The over-expression of SPINK1 was not only related to poor PSA response, but also significantly associated with the occurrence of CRPC, especially in those with much more aggressive phenotype (GS 8-10). So, SPINK1 could be considered as a useful prognostic predictor for bone metastatic PCa at the time of diagnosis, and further prospective studies are needed to verify the conclusions. Prostate 76:823-833, 2016. © 2016 Wiley Periodicals, Inc. © 2016 Wiley Periodicals, Inc.