Using network-time-protocol, NTP, to synchronize electronics has its limitations. Even if the NTP-server is a Stratum 1 operated by a responsible agent and connected to a reliable source, the asymmetries in the network and the influence by data traffic congestions limit the accuracy to the order of 10 ms or worse. The large uncertainty of a single request can be compensated by repetition, but to be useful the local oscillator must be more stable than the occurrence of congestions in the network. Even then, there is a remaining doubt on the accuracy of the time received by the client. In principle the time achieved from an NTP-server is to be seen as unreliable for legal purposes, if no feedback estimates of the uncertainties are acquired of the transmission. Experimental data present how this limitation has improved, at least statistically, through the continuous upgrade of data bandwidth in one of Sweden's back-bone networks. Even though data over Internet also increases, it appears that the detrimental effects of the network on NTP accuracy have decreased. Since 2011, SP have monitored a number of NTP-servers operated by TeliaSonera and logged the response compared to the realization of the national timescale in Sweden (UTC(SP)). All NTP-servers used in this evaluation are commercial, off-the-shelf products with stratum 1 synchronization. The reference NTP-server and NTP-logger are located in SP premises in Borås, with a virtual private network connection to all servers, and thus sharing the bandwidth of the network between server and client. The NTP-logger is customized and includes equipment for independent measurements of the round-trip-delay. The uncertainty of each server, as achieved when polled from Borås, is calculated. The results present how the precision improves with an order of magnitude during the last three years by evaluating maximum daily variations. The improvements during the years have decreased the uncertainty and enhanced the accuracy to stable levels in ms to sub-ms range. This paper will further study and present the performance of the NTP-monitoring, with reported offset and uncertainty. It serves as an empirical reference on network improvements enabling better performance of NTP-synchronization.