I had a unit that was 55 + Hours old and about 7% and I rebooted the PC without exiting from BOINC. When the PC restarted it did a benchmark and all the units were gone.

2004-08-30 22:16:40 [---] Starting BOINC client version 4.05 for windows_intelx86
2004-08-30 22:16:40 [climateprediction.net] Project prefs: using your defaults
2004-08-30 22:16:40 [Predictor@home] Project prefs: using your defaults
2004-08-30 22:16:40 [SETI@home] Project prefs: using your defaults
2004-08-30 22:16:40 [climateprediction.net] Host ID not assigned yet
2004-08-30 22:16:40 [Predictor@home] Host ID not assigned yet
2004-08-30 22:16:40 [SETI@home] Host ID not assigned yet
2004-08-30 22:16:40 [---] General prefs: from SETI@home (last modified 2004-08-27 22:15:41)
2004-08-30 22:16:40 [---] General prefs: using your defaults
2004-08-30 22:16:41 [---] Running CPU benchmarks
2004-08-30 22:16:41 [---] Suspending computation and network activity - running CPU benchmarks
2004-08-30 22:17:42 [---] Benchmark results:
2004-08-30 22:17:42 [---] Number of CPUs: 1
2004-08-30 22:17:42 [---] 958 double precision MIPS (Whetstone) per CPU
2004-08-30 22:17:42 [---] 2135 integer MIPS (Dhrystone) per CPU
2004-08-30 22:17:42 [---] Finished CPU benchmarks
2004-08-30 22:17:43 [---] Resuming computation and network activity
2004-08-30 22:17:43 [---] Insufficient work; requesting more
2004-08-30 22:17:55 [---] Insufficient work; requesting more
2004-08-30 22:17:55 [climateprediction.net] Requesting 78545 seconds of work
2004-08-30 22:17:55 [climateprediction.net] Sending request to scheduler: http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi
2004-08-30 22:17:59 [---] Running CPU benchmarks
2004-08-30 22:17:59 [---] Suspending computation and network activity - running CPU benchmarks
2004-08-30 22:19:00 [---] Benchmark results:
2004-08-30 22:19:00 [---] Number of CPUs: 1
2004-08-30 22:19:00 [---] 1055 double precision MIPS (Whetstone) per CPU
2004-08-30 22:19:00 [---] 2207 integer MIPS (Dhrystone) per CPU
2004-08-30 22:19:00 [---] Finished CPU benchmarks
2004-08-30 22:19:01 [---] Resuming computation and network activity
2004-08-30 22:19:01 [climateprediction.net] Scheduler RPC to http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi succeeded
2004-08-30 22:19:01 [climateprediction.net] General preferences have been updated
2004-08-30 22:19:01 [---] General prefs: from climateprediction.net (last modified 2004-08-30 19:38:34)
2004-08-30 22:19:01 [---] General prefs: using separate prefs for home
2004-08-30 22:19:01 [climateprediction.net] Project prefs: using separate prefs for home
2004-08-30 22:19:01 [Predictor@home] Sending request to scheduler: http://predictor.scripps.edu/predictor_cgi/cgi
2004-08-30 22:19:02 [climateprediction.net] File projects\\climateprediction.net\\hadsm3_4.03_windows_intelx86.exe exists already, skipping download
2004-08-30 22:19:03 [climateprediction.net] File projects\\climateprediction.net\\hadsm3data_4.03_windows_intelx86.zip exists already, skipping download
2004-08-30 22:19:04 [climateprediction.net] File projects\\climateprediction.net\\hadsm3um_4.03_windows_intelx86.zip exists already, skipping download
2004-08-30 22:19:05 [climateprediction.net] File projects\\climateprediction.net\\hadsm3se_4.03_windows_intelx86.zip exists already, skipping download
2004-08-30 22:19:05 [climateprediction.net] Started download of 01o9_400027154.zip
2004-08-30 22:19:07 [Predictor@home] Scheduler RPC to http://predictor.scripps.edu/predictor_cgi/cgi succeeded
2004-08-30 22:19:07 [Predictor@home] Message from server: To participate in this project, you must use major version 3 of the BOINC core client. Your core client is major version 4.
2004-08-30 22:19:07 [SETI@home] Sending request to scheduler: http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi
2004-08-30 22:19:07 [climateprediction.net] Finished download of 01o9_400027154.zip
2004-08-30 22:19:07 [climateprediction.net] Throughput 4559 bytes/sec
2004-08-30 22:19:07 [climateprediction.net] Starting result 01o9_400027154_0 using hadsm3 version 4.03
2004-08-30 22:19:09 [SETI@home] Scheduler RPC to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi succeeded
2004-08-30 22:19:09 [SETI@home] Project prefs: using separate prefs for home
2004-08-30 22:19:10 [SETI@home] File projects\\setiathome.berkeley.edu\\setiathome_4.03_windows_intelx86.exe exists already, skipping download
2004-08-30 22:19:10 [SETI@home] File projects\\setiathome.berkeley.edu\\better_banner.jpg exists already, skipping download
2004-08-30 22:19:12 [SETI@home] File projects\\setiathome.berkeley.edu\\setiathome_4.03_windows_intelx86.pdb exists already, skipping download
2004-08-30 22:19:12 [SETI@home] Started download of 04my04aa.2277.13009.1028410.222
2004-08-30 22:19:12 [SETI@home] Started download of 25ap04aa.185.22848.1034646.177
2004-08-30 22:19:16 [SETI@home] Finished download of 04my04aa.2277.13009.1028410.222
2004-08-30 22:19:16 [SETI@home] Throughput 86270 bytes/sec
2004-08-30 22:19:16 [SETI@home] Started download of 04my04aa.2277.11648.15918.214
2004-08-30 22:19:17 [SETI@home] Finished download of 25ap04aa.185.22848.1034646.177
2004-08-30 22:19:17 [SETI@home] Throughput 83744 bytes/sec
2004-08-30 22:19:17 [SETI@home] Started download of 25ap04aa.23553.17970.367328.176
2004-08-30 22:19:17 [climateprediction.net] Preempting result 01o9_400027154_0 (quit)
2004-08-30 22:19:18 [SETI@home] Starting result 04my04aa.2277.13009.1028410.222_4 using setiathome version 4.03
2004-08-30 22:19:21 [SETI@home] Finished download of 04my04aa.2277.11648.15918.214
2004-08-30 22:19:21 [SETI@home] Throughput 81739 bytes/sec
2004-08-30 22:19:21 [SETI@home] Finished download of 25ap04aa.23553.17970.367328.176
2004-08-30 22:19:21 [SETI@home] Throughput 110150 bytes/sec
2004-08-30 22:19:21 [SETI@home] Started download of 04my04aa.2277.11698.704834.120
2004-08-30 22:19:26 [SETI@home] Finished download of 04my04aa.2277.11698.704834.120
2004-08-30 22:19:26 [SETI@home] Throughput 75741 bytes/sec
2004-08-30 22:19:26 [climateprediction.net] Restarting result 01o9_400027154_0 using hadsm3 version 4.03
2004-08-30 22:19:26 [SETI@home] Preempting result 04my04aa.2277.13009.1028410.222_4 (quit)

This problem is not new. Sounds like BOINC might not have closed the model down cleanly. This bug were not fixed in BOINC v4.05. They are a 2 file rotation for the client_state.xml file in v4.xx which should solve alot of the lost workunit issues at restarts and crashes but it's not perfect. I already complain about this in the beta test.

If your want to avoid this, you may run BOINC as a service (I never get this problem) or you if you run GUI, you must Suspend computation before quitting application.