Recent studies have shown that the identification of deregulated bio-molecular pathways in cancer may be more important than identification of individual genes through differential expression. Since the same pathway can be deregulated by a different subset of genes, it is critical to study pathways as a whole, rather than focus on individual genes. Most papers on meta-analysis of cancer datasets focus on the identification of a set of individual genes, and not on pathways, protein families and gene ontology terms. We have analyzed data from 87 microarray datasets consisting of 5,126 samples and spanning 25 different types of cancer. We have identified 212 KEGG pathways, 578 protein families and 1,717 gene ontology terms that are statistically significant (p < 0.01), and deregulated in cancer. Many of the top pathways and proteins from our meta-analysis, e.g., Jak-Stat pathway, Annexin proteins etc. are already known to play a pivotal role in carcinogenesis. To the best of our knowledge this is the largest meta analysis of cancer pathways, protein families and gene ontology terms to date.