You don't say how you're measuring but I'll guess that you're using quantitative PCR. A typical approach to handling housekeeping gene normalization in qPCR experiments is described in
Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes.
We outline a robust and innovative strategy to identify the most stably expressed control genes in a given set of tissues, and to determine the minimum number of genes required to calculate a reliable normalization factor. We have evaluated ten housekeeping genes from different abundance and functional classes in various human tissues, and demonstrated that the conventional use of a single gene for normalization leads to relatively large errors in a significant proportion of samples tested. The geometric mean of multiple carefully selected housekeeping genes was validated as an accurate normalization factor by analyzing publicly available microarray data.
There is an R package called NormqPCR that implements this approach (publication here).
Briefly, the approach required measuring multiple housekeeping genes and iteratively removing the ones that are most variable, ending up with two genes that are jointly least variable, and making a sort of pseudo-housekeeping gene from them with which to normalize your other genes.
This doesn't magically turn bad data into good data, but it helps clean up some intrinsic problems with housekeeping genes.