Three UK studies on the relationship between a purpose-built instrument to assess the importance and development of 15 "soft skills" are reported. "Study 1" (N = 444) identified strong latent components underlying these soft skills, such that differences "between-skills" were over-shadowed by differences "between-students". Importance and improving ratings on these skills predicted academic performance and accounted for the effects of personality on academic performance. "Study 2" replicated the structure of the soft skills inventory and associations with academic performance in a larger sample (N = 1309). Examination of mean differences across faculties (humanities, life sciences, hard sciences) revealed higher soft skills ratings in "softer" courses. "Study 3" (N = 87) incorporated an IQ measure, which was found to be negatively related to importance ratings on soft skills. Results highlight the cohesive structure of beliefs concerning various non-academic skills and their significant links to educationally relevant individual differences. Theoretical, methodological and applied implications are considered. (Contains 7 tables, 4 figures and 1 note.)