We propose a sufficiently simple multichain model of ordered β-sheets, composed of extended macromolecules with rigid elements. The effective constants of intra- and interchain interactions describe primary and secondary structures of proteins, respectively. It is found that the long-range correlation of orientations of chain elements decreases with the separation along the same chain or between different chains according to the same asymptotic power law. The exponent in this law is determined by the ratio of the energy of thermal motion and the geometric mean of the energies of intra- and interchain interactions. The characteristic scale parameters are obtained, which define the crossover of the intra- and interchain correlation functions from the exponential law of decrease to the power one. The power law for intrachain correlations leads to a non-Gaussian behavior of the mean-square dimensions of chains. Several types of asymptotic dependences of mean-square dimensions of a chain in the β-sheet on the number of chain elements are found. Peptide chains may exist in different conformations: from extended ones to random Gaussian coils. Long-scale statistical properties of polymer systems with interchain interactions and those for polymer chains with excluded volume effects are compared.