A search is described for 3He, 4He, and tritium produced when D2 is absorbed by titanium sponge, or released when titanium deuteride is heated. The D2 is prepared from pre-nuclear-era D2O, which has a tritium/deuterium (T/D) ratio of 1.8 × 10−15. Two reservoirs of titanium sponge in a vacuum system attached to the inlet line of a mass spectrometer are heated to allow rapid transfer of D2 from one sponge to the other. Significant amounts of 3He and 4He are released only after the deuterium content is increased to reach TiD1.5 in one sponge. Then 3He and 4He are decreased as the D2 is transferred back and forth. When the titanium is loaded to a composition of TiD2.0, 3He and 4He increase during the next two transfers, then decrease. When the D2 is replaced by H2, then D2-H2 (1:1), 3He and 4He decrease steadily, indicating that the transfer process causes partial release of 3He and 4He trapped in the titanium. This view is supported by the fact that all fractions appear to have a constant 3He/4He ratio of 3.0 × 10−7. We believe that this helium is introduced from the cover gas used during the manufacture of the titanium sponge and that it has nothing to do with cold fusion. Assuming that the appropriate time is the transfer time of ∼1 h, the following upper limits are calculated: 1.4 × 10−21 fusion/d-d · s−1 for d + d = 3He + n, and 2.0 × 10−15 fusion/d-d · s−1 for d + d = 4He. The limit for the 3He channel is in agreement with the value of 10−23 fusion/d-d · s−1. After a series of transfers, the D2 is sealed in a container made of low-helium-permeability glass. After a decay time of 1.5 yr, tritium is assayed by measurement of 3He. The T/D ratio is found to be 6.4 × 10−15, significantly higher than T/D in the D2O. At present, because the possibility of tritium contamination cannot be eliminated, the excess tritium is viewed as an upper limit f or production by cold fusion. Assuming that the appropriate time is the total transfer time of 16 h, an upper limit is obtained for d + d = t + p of 1.6 × 10−19 fusion/d-d · s−1. Assuming that the appropriate time is the time D2 was resident in either titanium sponge, 360 h, the upper limit is 7 × 10−21 fusion/d-d · s−1. These limits are not in agreement with a rate of ∼10−14 fusion/d-d · s−1.