Because intelligence and culture are inextricably intertwined, the measurement of ability is difficult. Test material inevitably reflects the culture of the test developer. As a consequence of this culture-boundedness, a test might not measure the same thing in another culture, and scores might not be comparable across cultures. There are three main types of test incomparability (or bias): construct, item, and predictive. From the test user's point of view, the most important is predictive bias, for this has a direct bearing on processes of selection or placement of individuals in an organization. Predictive bias and fairness are not synonymous because fairness rests in the use of an assessment instrument and not in the instrument itself. There is no universally agreed-on conception of what is fair and what is unfair in an organization's relations with employees or prospective employees. It is incumbent on each organization to develop a clear fairness policy with regard to recruitment, selection, placement, and job advancement and to develop practical procedures to implement this policy. Various fairness models relevant to selection are discussed. It is recommended that organizations that do not have the skills to frame and implement a fairness policy either acquire these skills or engage the services of a consultant. (Author/SLD)