This study investigated mortality risks from all causes, circulatory and respiratory diseases for the elderly associated with prolonged exposure to extreme temperatures in four major cities of Taiwan. Daily average temperatures at the high 99th, 97th, and 95th percentiles were defined as extreme heat, and those at the low 10th, 5th, and 1st percentiles were defined as extreme cold for each city in 1994-2007. Distributed lag non-linear model was used to estimate the relative risk (RR) of mortality associated with 30-day lag temperature, and heat and cold extremes lasting for 3-5, 6-8, and >8 days. The random-effects meta-analysis summarized the risks of temperature and extreme temperatures events. The lowest overall mortality among the elderly was when the temperature was 26 °C on average. Low temperatures caused greater adverse effects than high temperatures, particularly for mortality from circulatory diseases. After accounting for the cumulative 30-day temperature effects, meta-analysis showed that mortality risk slightly increased with strengthened and prolonged heat extremes (≥99th and >3 days; ≥97th and >8 days; and ≥95th and >8 days) that RRs ranged from 1.04-1.05, 1.01-1.05, and 1.05-1.13 for mortality from all causes and from circulatory and respiratory diseases, respectively. The corresponding RRs ranged from 0.98-1.01, 0.92-1.06, and 0.97-1.03, respectively, for shorter duration of heat extremes. This study did not identify significant effect for stronger or prolonged cold extremes. Extreme temperatures and their duration cause varied mortality associations in the elderly. Short-term extremely low temperatures exhibit the greatest effect on mortality, and intensified and longer periods of heat extremes also exert a slightly increased effect on mortality. Copyright © 2011 Elsevier Inc. All rights reserved.