This dissertation is comprised of two peer-reviewed articles and one manuscript, aiming to identify knowledge gaps in terms of the role of telomere length as a biomarker of cancer. Individuals with relatively short telomeres in peripheral blood cells have been shown in epidemiological studies to be at increased risk for mortality and cancer, leading investigators to hypothesize that telomere length and attrition over the life-course is a critical mechanism underlying cancer and many other age-related health conditions. However, such associations are not consistent across all cancers or even within cancer types, with some studies showing null, or even opposite associations. Furthermore, due to the cross-sectional nature of case-control studies, from which most associations of telomere length and cancer risk are obtained, it is possible that telomere shortening occurs after diagnosis (due to the cancer itself or treatment effects), or is confounded by other common risk factors, and therefore may not be a true biomarker of subsequent cancer risk. I employed three strategies for addressing these potential issues: 1) use genetic polymorphisms that influence telomere length as proxies for telomere length to estimate associations between telomere length and cancer risk; 2) perform careful adjustment of potential confounders of the telomere length-cancer association, including the assessment of time-varying effects on smoking, a potential telomere length correlate; and 3) utilize a prospective study design in which baseline telomere length is measured prior to the development of any adverse health outcomes. In the first study of this dissertation (Chapter 2), I used a Mendelian randomization approach to estimate the associations between nine telomere length-associated SNPs and risk for five common cancer types (breast, lung, colorectal, ovarian and prostate cancer, including subtypes) using data on 51,725 cases and 62,035 controls. The long TL genetic score was significantly associated with increased risk of lung adenocarcinoma (P=6.3E-15), with an estimated odds ratio of 2.78. Under Mendelian randomization assumptions, this estimate was interpreted as the odds ratio for lung adenocarcinoma corresponding to a 1000 base pair increase in telomere length. The SNP score was not associated with other cancer types or subtypes. In the second study of this dissertation (Chapter 3), I estimated the associations between telomere length (measured by quantitative PCR using saliva DNA) and concurrent and past smoking status, reported biennially for up to 16 years prior to telomere length measurement, using data from the Health and Retirement Study (n=5,624). Smoking was associated with shorter telomere length when using prospective data on smoking status among men and women, but the association was strongly attenuated in cross-sectional analyses among men. This attenuation was largely due to a higher rate of smoking cessation during the study period among males with shorter telomere length compared to males with longer telomere length. Short telomere length was also associated with poorer overall health in men, suggesting that male smokers with short telomere length were more likely to quit smoking due to poor health. Analyses of years since cessation, smoking duration, and pack-years all support the hypothesis that increased cigarette use shortens telomere length. In the third study of this dissertation (Chapter 4), I investigated the association between arsenic exposure and telomere length, and the association between telomere length and incident arsenical skin lesion in the Health Effects of Arsenic Longitudinal Study (HEALS) in Araihazar, Bangladesh. In this prospective study, baseline telomere length was assessed in skin lesion-free subjects who were followed for up to 12 years. No association was observed between baseline arsenic exposure and telomere length. However, we observed higher incident skin lesion risk with shorter telomere length (Ptrend=4.6E-5), with odds ratios of 3.05, 1.30, and 1.21 for the first (shortest), second, and third telomere length quartiles compared to the longest. The role of telomere length as a biomarker of cancer appears to be striking yet nuanced. Addressing the existing knowledge gaps is a critical step towards clarifying the causal relationship between telomere length and cancer, and ultimately, improving cancer prediction and prevention.