DNA methylation (DNAm) based estimators of telomere length (TL) have been shown to better correlate with TL associated factors/outcomes than measured TL itself. Here, we examined the DNAm based estimator of TL (DNAmTL), the principal component version of DNAmTL (PCDNAmTL), qPCR measured TL (qPCR-TL), and southern blot measured TL in the same samples from a cohort from metropolitan Cebu, Philippines. In a pre-registered analysis, we examined the association between paternal age at conception (PAC) and offspring TL to better understand the meanings and utilities of PCDNAmTL and DNAmTL. We predicted that the nature of PCDNAmTL markings should determine whether they change in sperm with age and are passed on to the zygote like TL is. We failed to find an association between PAC and PCDNAmTL, but did find a positive association of paternal grandfather’s age at father’s conception predicting grandchild’s PCDNAmTL. Surprisingly, on almost all measures of external validity (correlations with parental TLs, southern blot TL, and age) qPCR-TL outperformed PCDNAmTL and DNAmTL. qPCR-TL showed a significantly stronger association with maternal qPCR-TL than PCDNAmTL and DNAmTL did. Similarly, qPCR-TL showed a significantly greater association with southern blot TL than PCDNAmTL or DNAmTL and the “kilobase” units of DNAm-based estimators of TL showed considerable deviations from southern blot derived kilobase measures. Also, surprisingly, for all of the measures of external validity DNAmTL outperformed PCDNAmTL. Overall, these findings suggest that PCDNAmTL and DNAmTL are not reliable indices of inherited aspects of TL and underscore uncertainty about the biological meaning of these measures.