Narrative:

ILS 24 bdl; vref=131; vtarget=136: autopilot engaged; leveled at 2;100 ft MSL; localizer captured; app mode armed; tracking and stable. Selected flaps 40 from 15 at 150 KTS and one half dot below glideslope. Glideslope captured and speed started to decay. Pilot flying announced; 'correcting airspeed;' and added power. Pilot not flying announced 'airspeed' as airspeed decayed through vref. Aircraft was still [indicating] on glideslope with airspeed decaying rapidly. Pilot flying applied maximum continuous thrust (mct). Airspeed decayed to 120 KTS; and the aircraft was still [indicating] on glideslope. Pilot flying pushed yoke forward into cws [control wheel steering autopilot mode]. At that moment the glideslope indicator quickly descended to the bottom of the glass. Altitude was 2;500 ft [400 ft above our altitude at glideslope intercept]. A go around was performed and we were vectored around for the same approach with an identical set up except that 2;500 ft MSL was maintained until glideslope capture. This time we briefed our plan of action to override the autopilot immediately if it attempted to climb. At glideslope intercept; the aircraft began climbing at 1;000 FPM and I pushed over into cws. Once in cws; the glideslope indicator began working properly again and dropped to one half dot high. We then rejoined the glide slope and landed uneventfully. I spoke to dispatch and maintenance control and made a logbook entry.the worst part of this event is that the display continued to show the aircraft on glideslope until knocked out of approach mode by pushing over into cws. The situation evolved rapidly from mind numbingly normal to something I have never seen before; appearing to be descending on glideslope at mct yet losing airspeed [because we were actually climbing]. The threat of continued loss of airspeed forced me out of believing the instruments into saving the aircraft from a potential stall.

Google
 

Original NASA ASRS Text

Title: Level; on the localizer and expecting a routine glideslope capture followed by an autopilot controlled ILS approach; the flight crew of a B737-300 was surprised and momentarily disoriented when the airspeed began to decay rapidly despite thrust increases as high as max continuous and indications the aircraft was descending on the glideslope.

Narrative: ILS 24 BDL; Vref=131; Vtarget=136: Autopilot engaged; leveled at 2;100 FT MSL; LOC captured; APP mode armed; tracking and stable. Selected flaps 40 from 15 at 150 KTS and one half dot below glideslope. Glideslope captured and speed started to decay. Pilot flying announced; 'Correcting airspeed;' and added power. Pilot not flying announced 'airspeed' as airspeed decayed through Vref. Aircraft was still [indicating] on glideslope with airspeed decaying rapidly. Pilot flying applied Maximum Continuous Thrust (MCT). Airspeed decayed to 120 KTS; and the aircraft was still [indicating] on glideslope. Pilot flying pushed yoke forward into CWS [Control Wheel Steering autopilot mode]. At that moment the glideslope indicator quickly descended to the bottom of the glass. Altitude was 2;500 FT [400 FT above our altitude at glideslope intercept]. A go around was performed and we were vectored around for the same approach with an identical set up except that 2;500 FT MSL was maintained until glideslope capture. This time we briefed our plan of action to override the autopilot immediately if it attempted to climb. At glideslope intercept; the aircraft began climbing at 1;000 FPM and I pushed over into CWS. Once in CWS; the glideslope indicator began working properly again and dropped to one half dot high. We then rejoined the glide slope and landed uneventfully. I spoke to Dispatch and Maintenance Control and made a logbook entry.The worst part of this event is that the display continued to show the aircraft on glideslope until knocked out of approach mode by pushing over into CWS. The situation evolved rapidly from mind numbingly normal to something I have never seen before; appearing to be descending on glideslope at MCT yet losing airspeed [because we were actually climbing]. The threat of continued loss of airspeed forced me out of believing the instruments into saving the aircraft from a potential stall.

Data retrieved from NASA's ASRS site as of July 2013 and automatically converted to unabbreviated mixed upper/lowercase text. This report is for informational purposes with no guarantee of accuracy. See NASA's ASRS site for official report.