A two-month aerial operation dropping 1080 on 43,000ha of the island finished on August 22.
The aim of the operation was to kill feral cats endangering the southern dotterel(pukunui) population on the island.
It was expected the cats would feed on the carcasses of rats and possums killed by the bait.
In October and November, two dead kiwi were found outside the area where the bait had been dropped.
The carcasses were frozen and sent to Manaaki Whenua Landcare Research Group for testing.
In the method of testing used, no amounts of 1080 were found in the muscle samples.
Doc acting operations manager Grace Tocker said the Manaaki Whenua tests showed the kiwi did not have 1080 in their system.
"In addition, monitoring before and after the predator control operation showed no change in kiwi numbers in the area after 1080 was applied."
The poison could not have degraded in the birds from the time they were discovered and the time the testing was done because the carcasses had been frozen, she said.
It was possible one bird could have died in the operational area and been swept downstream to where it was found but the other bird was found a long way from the area where the 1080 had been dropped, she said.
It was not known how long the birds had been dead before they were found, she said.
Decades of research showed the risk of kiwi being killed by 1080 was small, she said.
Since 1990, about 800 radio-tagged kiwi nationwide have been monitored through aerial and ground-based 1080 predator control operations.
"In that time, only two kiwi have ever been found with 1080 residue and in both cases, the cause of death could not be confirmed."
The latest estimate showed there were about 20,000 adult kiwi on the island and it was not uncommon for people to find dead kiwi on the island.
Since 2008, 35 dead kiwi have been handed into or reported to Doc on Stewart Island.
Most had been found near Oban township and the likely causes of death were drowning, starvation, dog attacks or being hit by a car.
This did not include the kiwi that died of natural causes throughout the national park, she said.
A Manaaki Whenua Landcare Research Group Bioeconomy Science Institute spokesperson said muscle was tested, not the stomach contents, because of the length of time the kiwi had been dead.
"Stomach contents are best for fresh samples as they will show the highest levels initially, but 1080 is rapidly absorbed from the stomach into the bloodstream and then muscle becomes the better sample to test."
The method used to test the tissue can identify 1 thousand-millionth of a gram of chemical per gram.
"If more than 1 thousand-millionth of a gram of 1080 had been present in any 1g of tissue, the test would have detected it, which is a very high level of precision."
