We simulate the anisotropy in the cosmic microwave background (CMB) induced by cosmic strings. By numerically evolving a network of cosmic strings we generate full-sky CMB temperature anisotropy maps. Based on 192 maps, we compute the anisotropy power spectrum for multipole moments ᐉ # 20. By comparing with the observed temperature anisotropy, we set the normalization for the cosmic string mass per unit length m, Cosmic strings are topological defects which may have formed in the very early Universe and may be responsible for the formation of large-scale structure observed in the Universe today [1]. In order to test the hypothesis that the inhomogeneities in our Universe were induced by cosmic strings one must compare observations of our Universe with the predictions of the cosmic string model. This Letter presents results of detailed computations of the large angular scale cosmic microwave background (CMB) anisotropies induced by cosmic strings [2]. These predictions are compared to the large-scale anisotropies observed by the Cosmic Background Explorer (COBE) satellite. Because the predicted temperature perturbations are proportional to the dimensionless quantity Gm͞c 2 where G is Newton's constant and c the speed of light, one may constrain the value of m, the mass per unit length of the cosmic strings. We believe that our estimate of m is the most accurate and reliable to date.Our methodology for computing the large angle anisotropy is to simulate the evolution of random realizations of a cosmic string network [3]. From these network simulations we construct the temperature anisotropy pattern seen by various observers within the simulation volume. We have evolved the strings from a redshift z 100 to the present, in a cubical box whose side length is twice the Hubble radius at the end of the simulation. This large box assures us that the anisotropy pattern is unaffected by the finite simulation volume.In order to obtain the large dynamic range required for these simulations we have used a new technique whereby the number of segments used to represent the string network decreases as the simulation proceeds. We have conducted tests of this method by comparing smaller simulations, with and without decreasing the number of segments: the average long string energy density is unaffected; the distribution of coherent velocities (the string velocity averaged over a particular length scale) is preserved down to scales smaller than 1͞100 of the horizon radius; the effective mass per unit length of string (the energy in string averaged over a particular length scale) is preserved down to scales smaller than 1͞100 of the horizon radius. The decrease in the number of segments was regulated so that on the angular scales of interest the simulation provided a good representation of the cosmic string network. Here we are interested in comparing with data from the COBE differential microwave radiometer (DMR) which measures the anisotropy convolved with an approximately 7 ± FWHM beam [4]. Our contact 0031-9007͞96͞77(15)͞3061(5)...