That depends on what you mean by reliable. It is common to be able to get dates that roughly match the evolutionary timescale. The published literature mostly reports dates which match that timescale. However, it is also common to get dates that do not match the evolutionary timescale. The exact proportion of "accurate" to "inaccurate" dates is difficult to ascertain from the literature.
Two factors are involved. First, in a given rock, for example lava, at least 4 dates can be obtained: the whole rock, and three fractions (light density, heavy density magnetic, and heavy density nonmagnetic). These can give dates with the oldest being over 4 times the youngest, on the same rock. (1)
Second, some dates are not reported, because their results are not "acceptable" . One paper on geochroknology mentioned a research project in which approximately 85% of the data were never published. (2)
References:1. P. E. Damon, A. W. Laughlin, and J. K. Percious ("Problem of excess argon-40 in volcanic rocks." In: Radioactive Dating and Low-level counting. Vienna:IAEA, 1967, pp. 463-481) note several such instances, including one case where olivine in a recent (<13,000 year old) lava dates over 110 million years old. Steven A. Austin (Excess argon within mineral concentrates from the new dacite lava dome at Mount St. Helens Volcano, Creation Ex Nihilo Tech J, 10:3 (1996), pp. 335-343 notes lava from recent flows at Mount St. Helens to date from 340,000 years to 2.8 million years, depending on the rock fraction. 2. J. F. Evernden, D. E. Savage, G. H. Curtis, and G. T. James ("Potassium-argon Dates and the Cenozoic Mammalian Chronology of North America." Am J Sci 1964; 262:145-198) report on p. 174, "Unfortunately many of the samples that passed field inspection for suitability and were laboriously collected, later proved unsuitable for dating. . . . Thus of some 65 samples collected by M. Skinner, only 10 could be used. This was also true of tuffs from the type Clarendonian sent to us by W. Chamberlain. H. A. Powers and D. Taylor also collected and painstakingly concentrated a large number of specimens that could not be used." Thus 85% of the samples submitted by Skinner were not reported. His is not an isolated case. It may be argued that this was early in the history of potassium-argon dating, and that the field has advanced since this article. This may very well be true. However, this article was considered a classic, and is still used by some who believe in long ages to prove their point. It also contains one of the largest series of independent dates.
2010 Arthur V. Chadwick, Ph.D.