This paper examines whether the clustering of initial public offering (IPO) spreads in the IPO market occurs through competition or collusion, using the period from January 2000 to July 2006 as the analytical range. It finds that the Korean IPO market spread is clustered at 3.5% on moderately sized IPOs (i.e., IPOs between 9 (KRW billion) and 33 (KRW billion)), representing a 74% overall rate that increases gradually over time. Our empirical results refute the collusion theory. The Korean IPO market is characterised by low concentration. A 3.5% spread does not represent abnormal profits relative to other IPOs. Cost saving (i.e. an inverse U shape) is confirmed by the use of 3.5% offers. Our results also show that factors typically related to underwriters and issuers – such as proceeds, secondary sales, earnings and reputation – affect the 3.5% spread used both positively and negatively. Thus, consistent with the efficient contract theory, the use of a 3.5% contract becomes less likely as proceeds increase but becomes more likely for IPOs of over 35 (KRW billion). On the one hand, the use of a 3.5% contract becomes more likely the more insider shares an IPO contains and the higher the volatility of the stock return (an insignificant factor). On the other hand, the use of a 3.5% contract becomes less likely the higher the earnings and debt (insignificant factors) and the more reputable the underwriter. Underpricing in 3.5% IPOs is conditioned by proceeds and secondary sales, not reputation and IPO market concentration. Thus, we conclude – against the cartel theory – that there is no collusive conspiracy in the Korean IPO market and that underwriters do not reap excess profits from 3.5% IPOs.