What can make radiation monitor go crazy like this?

I just had the weirdest thing happen to me and my wife. I have a Radex 1503 counter that I bought mostly for fun measurements of rocks, etc, and for the unlikely event that there might ever be a nuclear accident near where I live. Recently I had a conversation with a friend from Japan about contamination in seafood and when doing some googling I came across reports that Korean seaweed and even Canadian salmon (which apparentaly migrates close to Japan) were being tested for contamination potentially resulting from the Fukushima incident.

Today I got 10 pounds of Alaskan sockeye salmon during a special deal at whole foods, and on the way home thought that it might be fun to see if there are even he slightest levels of radiation increase. So when I got home I put the counter on it and let it go through its measurement cycles. I took the counter off the salmon about ten minutes later, the counter registering normal background levels of 0.1-0.2 u Sv/ hour up until that point. Having confirmed my expectation that the salmon would be completely or at least immeasurably uncontaminated, I left the counter on the kitchen counter for a few minutes, having gotten distracted by some work stuff I needed to take care off. All of a sudden the counter started beeping (my wife alerted me to it) and measuring high levels of 9.99 uSv/hour (the maximum level) consistently. Clearly that had nothing to do with the salmon as it had been registering normal levels up until that point. However the counter continued registering these levels, fluctuating between 2.5 and 9.99 uSv/hour for about 10-15 minutes. At that point my wife and I decided to call the local police/fire department, be it just to hear their opinion. They deemed the situation was worth sending out a fire truck which arrived about ten minutes later. At that point my dosimeter was registering slightly lower levels on average, but was still fluctuating between 0.5 and 6 uSv/hour, definitely more than normal background levels. I greeted the firemen outside our house and explained the situation to them. They got out their own dosimeter (sensitive to gamma only, mine being sensitive to beta and gamma), and one of them went with me into the house for a minute measuring the levels. His dosimeter registered nothing unusual, so quite understandably they took off, probably thinking I'm some sort of crackpot. After they left I went back into the house and the device was again registering levels of 2.25 uSv/hour, or about twenty times above average. However the mean started getting lower. About twenty minutes later, the levels, measured in different parts of the house, where again at completely normal background levels, always around 0.1-0.2 uSv/hour.

To summarize, I'm somewhat embarassed yet very grateful to the fire department for coming out for this scare. The reason I'm posting here is to find out how it can be that a Geiger-Mueller counter from a respected company like this one can read such consistently wrong levels over such an extended measurement period?

Thanks in advance for any thoughts you might have.