Security firm Damballa published research this week suggesting that the average US user is more likely to be struck by lightning during their lifetime as they are to be infected by the most serious mobile malware in a particular year.
A fortnight ago Verizon reported being unable to find much malware on the devices of its millions of US users and Damballa's figures are even more emphatic if less detailed.
Using a dataset made up from monitoring half of all US mobile network traffic, the firm said that in the fourth quarter of 2014 it had spotted only 9,688 devices contacting malware domains out of a population of 151 million devices, a detection rate of 0.0064 percent.
The total number of domains contacted by this user base was 2,762,453.
This compared to the odds from the US Weather Service that assesses the lifetime risk of being hit by lightning as 0.01 percent, Damballla said.
"This research shows that mobile malware in the Unites States is very much like Ebola - harmful, but greatly over exaggerated, and contained to a limited percentage of the population that are engaging in behaviour that puts them at risk for infection," said Damballa scientific researcher, Charles Lever.
"Ask yourself, 'How many of you have been infected by mobile malware? How many of you know someone infected by mobile malware?'"
One question is whether Damballa is seeing all the mobile malware on these devices. Detecting it using this methodology requires identifying malicious domains accurately and it's possible some are not yet in the database.
If a domain isn't classified as malicious, command and control or traaffic sent via it won't be picked up either. There's also the issue as to what counts as 'malware'. Some security firms count almost anything, including spyware and adware, which is not included in Damballa's definition. That seems reasonable - many spy and adware programs are willingly installed by the user in return for being able to use a mobile app for free.
On the other hand, mobile malware can be lower-level than PC equivalents and still represent a hazard, for example apps that run up SMS toll fraud. Because many apps can make in-program charges, outwardly the app might look more harmless than it is.
"Mobile operators and platforms have invested significant resources in preventing malicious applications from being installed, especially in North America. For example, iOS developers must submit an application for approval before their app is available on iTunes.
"And Google has developed 'Bouncer,' a system that scans submitted apps for evidence of malware.
By staying within vendor app stores users greatly minimised their chances of picking up the nastiest malware, in the case of Apple to almost zero.
The lightning analogy strike is perhaps slightly misleading, comparing lifetime risk with the risk in a single year. The point remains that mobile malware is not as common as some have claimed it is, at least in the US.
Often estimates are based on samples found circulating rather than actual infections, which is not the same thing. Both Damballa and Verizon measured what they had found on real devices.