The linkage disequilibrium method is currently the most widely used single sample estimator of genetic effective population size. The commonly used software packages come with two options, referred to as the parametric and jackknife methods, for computing the associated confidence intervals. However, little is known on the coverage performance of these methods, and the published data suggest there may be some room for improvement. Here, we propose two new methods for generating confidence intervals and compare them with the two in current use through a simulation study. The new confidence interval methods tend to be conservative but outperform the existing methods for generating confidence intervals under certain circumstances, such as those that may be encountered when making estimates using large numbers of single-nucleotide polymorphisms.