The k-out-of-n systems with components having random weights have wide applications in reliability engineering. In this paper, we implement the study on reliability analysis of k-out-of-n systems with heterogeneous components and random weights. Firstly, for a randomly weighted k-out-of-n system with dependent components chosen from two different classes of ordered components, we carry out stochastic comparisons of different random selection strategies with respective to the usual stochastic order, the increasing convex [concave] order, and the expectation order. Secondly, we investigate the performance levels of randomly weighted k-out-of-n systems with components chosen via independent Bernoulli trials under different selection probabilities. Thirdly, we consider the assembly of a randomly weighted k-out-of-n system through drawing n subsystems from m groups with each group consisting of independent and identically distributed components. The effects of selecting strategies and sampling probabilities on the systems total capacities are studied according to the above mentioned stochastic orders when the subsystems possess parallel or series structure. These results can be conducive to assembling randomly weighted k-out-of-n systems in practical scenarios. Numerical examples are provided to illustrate the findings as well.