The first purpose of this paper is to shed some new light on the old question of selecting the number of beams in intensity-modulated radiation therapy (IMRT). The second purpose is to illuminate the related issue of discrete static beam angles versus rotational techniques, which has recently re-surfaced due to the advancement of volumetric modulated arc therapy (VMAT). A specific objective is to find analytical expressions that allow one to address the points raised above. To make the problem mathematically tractable, it is assumed that the depth dose is flat and that the lateral dose profile can be approximated by polynomials, specifically Chebyshev polynomials of the first kind, of finite degree. The application of methods known from image reconstruction then allows one to answer the first question above as follows: the required number of beams is determined by the maximum degree of the polynomials used in the approximation of the beam profiles, which is a measure of the dose variability. There is nothing to be gained by using more beams. In realistic cases, in which the variability of the lateral dose profile is restricted in several ways, the required number of beams is of the order of 10-20. The consequence of delivering the beams with a 'leaf sweep' technique during continuous rotation of the gantry, as in VMAT, is also derived in an analytical form. The main effect is that the beams fan out, but the effect near the axis of rotation is small. This result can serve as a theoretical justification of VMAT. Overall the analytical derivations in this paper, albeit based on strong simplifications, provide new insights into, and a deeper understanding of, the beam angle problem in IMRT. The decomposition of the beam profiles into well-behaved and easily deliverable smooth functions, such as Chebyshev polynomials, could be of general interest in IMRT treatment planning.